Bent Erik Bakken Models the World

Bent Erik Bakken thinks big. A graduate of the University of Lausanne with a degree in business administration, he could have followed his peers into finance and management, but he wanted to understand how the world works. “There was too much to hold in my head,” says Bakken. “What I wanted was a way to organize information, push a button, and see what happened.”

Bakken found that methodology 35 years ago when he discovered system dynamics while working at the Norwegian Defense Research Establishment. He went on to earn a PhD in System Dynamics from the Massachusetts Institute of Technology (MIT) and started a lifelong career in investigating complex systems through models and simulation.

For his doctoral dissertation, Bakken modeled the American real estate system, and his interactive learning sessions helped save a large firm from going bankrupt in the market upheaval of the late 1980s. When he returned to Norway, he developed resource allocation and combat simulation models. One model looked into the future of NATO’s effort in Afghanistan. “The model showed that success in Afghanistan was close to impossible,” says Bakken. His wide variety of interests also include how Oslo’s waste system can be improved, and he has worked with the city of Oslo identifying ways to increase recycling rates.

After 25 years in defense analysis, Bakken joined DNV eight years ago, a 12,000 people-strong Norway-based global society in assurance and risk management active in 100 countries, and the scope of his work expanded. “At DNV, our research team had the audacity to develop models that see the world as an interconnected system defined by the links within,” says Bakken. “The Future of Spaceship Earth report, was developed in close cooperation with Jørgen Randers, one of the authors of The Limits to Growth.”

Published in 1972, The Limits to Growth explored exponential global expansion with a finite resource supply. Originally developed with DYNAMO and later revised and maintained with Stella®, the model underlying The Limits to Growth is considered seminal work in the field of system dynamics. The Future of Spaceship Earth continues to test the likelihood of achieving the United Nations’ 17 Sustainable Development Goals by the year 2030.

Now the head of his own company, Sustainability Dynamics, Bakken continues to service two other global-scale models. The Energy Transition Outlook (ETO) model forecasts the differential speed of the transtion from fossil fuels to renewable energy in ten global regions. The Ocean-Space Forecast models every aspect of the value of oceans, including shipping, tourism, fisheries, and energy extraction.

“ETO is instrumental to my client DNV and its stakeholders,” says Bakken. “Energy is at the core of life, society, and industry. The annual ETO report bases its projections on the models’ results. The report is downloaded 150,000 times each year, more than any other DNV report. DNV’s CEO, Board of Directors, and senior leaders use the insights to devise its own future strategy. Stakeholders outside the company, from students to energy analysts to government officials to business executives, read the report and download regional presentations of its findings.”

While Bakken recognizes that architecting global models is “not an exercise for the meek,” he makes it sound almost easy. “You have to understand which parts of a system are the same around the world and which differ. The energy system as a feedback loop, for example, is identical region to region, but you need to reflect also regional differences. The United States and Canada are similar in many ways, but both are very different from Latin America. The way some energy policies are created in the United States is very different from the way it’s created in sub-Saharan Africa.”

To tame complexity, Bakken coaches the DNV team to apply the Three S’s. “Simplify, Simplify, Simplify.” “I caution my team that we don’t end up with a model that we do not understand. Speaking to some US colleagues that lack a system dynamics background, they described their own model as a Beast. That is something we must avoid.”

The DNV teams use Stella Architect to continuously improve the Energy Transition Outlook and Ocean-Space Forecast models. “I think DNV holds about a dozen licenses and users,” says Bakken. “Flexibility is important to us as we have users who are IT professionals able to write code, and others who rely on Stella’s graphical interface. More important, we are system dynamicists, most of us with PhDs infused with a modelling approach, and value the high quality and dedication to system dynamics evident in Stella Architect.”

Each year, DNV models are updated with the most current energy and policy information, like electric vehicle sales and renewable energy system capacity. They also add systems elements as they become important on a global scale. “When we started, the Energy Transition Outlook model didn’t include hydrogen, but it became clear after two years that understanding hydrogen is key to understanding the speed of the transition,” says Bakken. “The model now includes six different ways to produce hydrogen, some fossil fuel-based and others based on bespoke and grid electricity.”

The ETO model soon reflected that the energy system will rapidly electrify using mostly intermittent solar and wind sources. Though somewhat counter to the need for simplification, Bakken’s team found it mandatory to reflect the interplay between intermittent power supply and various forms of storage and other flexibility options. “The model needed hourly resolution, though it typically runs seventy years from 1980 to 2050,” says Bakken. “Given the size of the model, it would have taken a week to run with that level of granularity, and our desktop computers would not have been able to do it. Onur Özgün, a really smart team member, realized that we only needed hourly resolution for the power sector of the model. He created an array of hours of the week within Stella and the model now runs in two hours on laptop computers.”

Though smaller models are helping Bakken’s clients explore local or regional systems, at Sustainability Dynamics, he is still very focused on global issues. “Every year we strive to understand better,” says Bakken. “Though, in 2022, I’m also pushing the team to remove some superfluous complexity.”

Modeling Mental Health System Improvements Gets a Pandemic Lift

Modeling and simulation have played important roles in understanding and explaining the physical health implications of COVID-19. Their insights are helping public health officials around the world project infection rates and understand the impact of interventions like masking, social distancing, and vaccinations. But what about the mental health impacts of COVID-19? How has the pandemic exacerbated mental illness risk factors? What actions can federal and local governments take to increase their communities' mental health?

“The Covid-19 pandemic has been a watershed moment for systems modelling in mental health,” says Jo-An Occhipinti, Associate Professor, Head of the Systems Modelling, Simulation, and Data Science department, and co-Director of the Mental Wealth Initiative at the Brain and Mind Centre at the University of Sydney. “Where there had been pushback from mental health epidemiologists who are more comfortable with traditional statistical methods, there is now a new level of systems modelling literacy. Models can drive a complex change process. They allow people with different agendas to learn together, which is an effective way of catalyzing coordinated action to improve community mental health outcomes.”

Along with her peers at the Brain and Mind Centre, A/Professor Occhipinti is working with Primary Health Networks (PHNs) across Australia to grapple with pandemic-related increases in mental illness and suicide risk, mental health service demand, substance use, and homelessness. In place of traditional statistical analyses that examine risk factors in historical data, models and simulation are helping regional stakeholders to bring together best available research evidence, expert knowledge, and data to understand the future trajectory of population mental health outcomes and evaluate interventions.

“While understanding individual risk factors is important, to make robust forward projections we need to understand the complex ways they interact, that’s where systems modelling plays an important role,” says A/Professor Occhipinti. Decades of research have identified individual, familial, social, economic, and environmental risk factors that represent possible targets for intervention, but, in the past, this has often resulted in action plans that include laundry lists of unrelated interventions that has made evaluation and accountability difficult.

To address these issues, the Brain and Mind Centre builds models of national, state, and regional mental health systems to find effective ways to reduce negative outcomes. “We use Stella® Architect to build models that not only capture the dynamics of the mental health system and its ability to respond to increases in demand, as seen during the pandemic, but also to understand how policies and initiatives to improve education and training and reduce unemployment, homelessness, and other social issues can impact mental health outcomes,” says A/Professor Occhipinti. With help from her and her colleagues, PHNs use Stella® Architect to use models customized for their regions to run simulations, and identify, explain, and evaluate intervention strategies.

“Stella Architect is the best invention!” says A/Professor Occhipinti. “You don’t have to be a software engineer to develop the model interfaces. It’s really easy for teams to build flexible interfaces that support on-going, inclusive use of models.”

When the pandemic hit, existing systems models were rapidly refined and deployed. PHNs could test which initiatives would mitigate the social and economic impacts of the pandemic on population mental health and decide what to fund, implement, and measure.

“Our models expose where the smart investments lie, which is so important for knowing how best to allocate limited federal and state funding,” says A/Professor Occhipinti. “Participatory approaches to systems modelling and simulation aided by Stella Architect also help us bring people on the journey of learning with us. It has helped us demonstrate important lessons like ‘more is not necessarily better,’ that sometimes a small set of interventions that work well together and are delivered at the right scale can be more effective than investing in too broad a range of programs and initiatives. Also, participatory systems modelling has helped stakeholders understand that sometimes, the most effective population mental health strategies lie outside of the health sector.”

For example, the Job Keeper program, implemented during the pandemic when unemployment rose significantly, kept workers connected to their workplaces and prevented further unemployment and the social dislocation that comes with it. Other programs aimed at increasing female employment and providing adequate childcare were projected to deliver good impacts in reducing mental health related emergency department presentations and self-harm hospitalizations in Australia.

PHNs have also learned that the timing of intervention implementation is critical. “We know that early intervention is key to reducing suicide rates,” says A/Professor Occhipinti. “There is good evidence that training general practitioners to recognize the signs of suicide ideation and refer to appropriate services as well as mental health education programs are effective in improving early care seeking. However, model projections have helped PHNs see that when these interventions are introduced in a region whose mental health system is already stretched, there are unintended consequences. As a result of increased waiting lists for specialist mental health services, people disengage from the system, and mental health outcomes get worse. PHNs have to address system constraints and capacity before other interventions can be effective.”

The economics of implementing interventions also impact the decisions made. The Brain and Mind Centre systems models include costs so that PHNs can apply different economic perspectives across alternative time horizons to understand the cost-effectiveness or cost-benefit associated with alternative strategies for funding allocation. “The flexibility afforded by the model interface allows us to provide the economic outputs with a level of transparency, highlighting for each intervention simulation who pays and who benefits,” says A/Professor Occhipinti. “For example, a program might be funded by the federal government, but state governments save money when hospitalizations are avoided. This often leads to fascinating discussions including where funding lobbying should be applied.”

The ability to create customizable user interfaces has kept Stella Architect front and center in the Brain and Mind Centre’s work. “How a model interface looks, and its usability, can really flavor the user’s perception of its value,” says A/Professor Occhipinti. “When end users are empowered to change its design, its functionality, and the language it uses, users make the model their own, which aids in sustainable use of the decision support tool. One PHN easily changed their model interface to reflect new branding which kept Stella Architect and the model out in the open.”

In October 2021, the research team at the Brain and Mind Centre published Beyond tinkering with the status quo: Coordinated structural reforms can deliver a substantive improvement in mental health outcomes. The report describes the group’s use of systems modeling to understand the likely impacts of key mental health system and service reforms. Specifically, the model points to technology-assisted care coordination between multiple service providers, a substantial increase in specialist mental health service capacity (psychiatrists, psychiatric nurses, occupational therapists, social workers, etc.), and enhanced direct access to care that eliminates wait times between system entry points and specialist care.

Importantly, the systems modeling supporting the report demonstrated that any of these three key elements on their own does not achieve a significant impact, and some may make the situation worse. However, when implemented in combination, modeled reforms were projected to deliver synergistic effects. The findings of the Brain and Mind Centre’s systems modeling program point to the importance of systems modeling beyond infectious disease control, to inform the design and strengthening of mental health and social systems and the allocation of limited resources to deliver better outcomes.

Capacity building is also an important next step for the field of system dynamics. As Managing Director of Computer Simulation and Advanced Research Technologies (CSART), an international alliance working to build global capacity in modeling and simulation to support health and social policy and planning decisions, A/Professor Occhipinti is leading efforts to share work across national boundaries. “Substantial technological and methodological advancements are being made by centres of excellence in health-focused systems modelling around the world, which are improving the accessibility, transparency, and robustness of these modelling tools,” says A/Professor Occhipinti. “It’s so important that we share these advances and put sophisticated decision support tools in the hands of decision makers so they can deliver more effective public policy and health systems and make direct and profound impacts on people’s lives.”

“Systems modelling remains a rare skill set in health, so we are also very keen to build much needed capacity to do this important work. Now that the pandemic has shown the benefits of models, we don’t want to fall back into old ways.”

Stella Users Network

If you have not yet, join our Stella Users Network group! Ask questions about our software, modeling, or systems thinking and receive answers and comments directly from us, as well as from a community of experienced modelers. Many Stella users are already members, and the more of us, the better it will be. We have been overjoyed with the feedback that questions and posts have received. To join, go here and click on Join This Group at the bottom. Should you have questions or issues about signing up, please email us.

COVID Update from Our Office

In the continued effort to prevent the spread of COVID-19, our office will remain closed. We will continue to work from home to offer our products and to support our customers. You may not always reach someone immediately by telephone, but we will continue to check voicemails and respond as quickly as possible. For more immediate assistance, email and chat will be available during our office hours (Monday through Friday, 9 am to 5 pm ET).

On the Road...Sort Of

As the pandemic persists, we continue to travel the virtual road.

In early November, Co-president Karim Chichakly attended the 19th annual Latin American Conference on System Dynamics. Karim presented his work on the implications human behavior has on containing and ending the pandemic. This presentation is available to view on our webinars page or through this link, and the accompanying article published in System Dynamics: Insights and Policy Innovations is available here.

In our spring Connector, we were happy to announce we were sponsoring the fourth annual South African Chapter System Dynamics Competition. Competitors used Stella Architect to model how COVID-19 has changed their area of interest. This year there were 50 competitors from 19 different countries. After months of instructions and modeling, the winner was announced this week during the South African Chapter Conference. Congratulations to Estee Vermuelen for her work on the impact that COVID-19 is having on the tourism industry of the Nelson Mandela Bay area.