Systems Innovation Practice’s experience in facilitating strategic insights covers a broad range of fields and applications. A sample of dynamic modeling projects we have completed in Health Care,
Business Strategy, and Energy, along with the key insights that were discovered along the way, appears below.
Pandemic patient management
The chief medical officer of a managed care organization was concerned about the potential impact on covered costs from a pandemic. The actuaries calculated a number of plan members who would become
infected and the cost of treating them, which they added to other covered costs to arrive at a startling new estimate for the year. In our group, we noted that overall capacity was tightly constrained
and most pandemic patients would either displace less severe cases or be untreated. Over the longer term, costs likely would decline.
Managing employee health care coverage
A pharmacy store-based nurse practitioner initiative sought to be added to health plan coverage, which would create a new provider group and headaches for claims processing systems. Marketing dearly
wanted to please its clients with coverage. The question was how the initiative would affect plan costs. We built a system dynamics model to simulate its effect, which indicated that most visits
would be for non-covered conditions and volume at each site would be too low to sustain the initiative. Coverage was granted but the initiative fizzled.
Management of patients with chronic disease
An ounce of prevention is worth a pound of cure. That adage has underlain many public health initiatives, and definitely applies to communicable diseases. As attention focuses on chronic conditions,
however, the relationship between treatment, health and cost becomes more subtle. A small model demonstrated that disease related mortality can apparently rise with better treatment. The seeming
contradiction is easily explained as a result of timing (death can be deferred, but rarely avoided). That insight is helpful in choosing metrics for program performance.
Organ transplant system in the US
A system dynamics model of the patient and service chains was created to experiment with various policies and practices in order to understand their impact on waiting times and long term impact on services,
developed as a part of a larger study on transplant organ potential. An important insight of the study was the recognition of policy resilience of the system. It furthermore helped to identify effective
measures that were initially under the radar screen.
Model-based Learning Environments
A model allows the consequence of policies and decisions to be studied quickly without impacting the organization. Creating a learning environment, also called a management flight simulator and in
essence a computer game, allows people to experiment with running things. Learning environments can be implemented as standalone experience for individuals or teams working together, or as part of
a larger group interaction where teams might be competing or might represent different functional areas within an organization. When combined with appropriate curriculum development, learning environments
can be highly effective at providing training in compressed time frames.
Managing stroke patient chains
A simple model of the patient chain for understanding the impact of various service improvement initiatives, developed for training purposes. One advantage of having a simple model was that people could
relate to it while it also lent itself easily to experimentation. Experimentation with the model showed that local process improvement efforts may create adverse affects for the whole chain.
Managing medical referrals
A specialty medical practice was having difficulty remaining profitable with referrals to the practice fluctuating over time. A system dynamics model was developed that captured the referral cycle and showed
the managing partner his referral base was too large and that too many referrals created a backlog of patients that increased waiting times to the point where patients and referring doctors went elsewhere.
The managing partner was able to use the model to convince the other partners to implement a strategy of controlling referral sources. Since the policy was implemented the practice has tripled in size and
Supply chain management without demand forecasting
Accurate demand forecasting is an essential part of most supply chain management algorithms. Yet, supply even in the most diligently managed chains can deviate widely from demand. Additionally,
any disturbance in demand will often be amplified in a supply chain creating what is known as bullwhip effect. This simulation attempts to demonstrate using three parsimonious models that supply
can be regulated reliably without forecasting demand. The inspiration for these models comes from PID (Proportional, Integral, Derivative) control in engineering controllers that utilize multiple
functions of error (discrepancy between goal and actual condition) to regulate a quantity.
Changing fleet maintenance strategies
A company wanting to improve the reliability of its fleet and decrease labor costs associated diagnosis and repair of failed systems was moving to a new more proactive maintenance practice.
They were concerned about the short and long term implications of this for its workforce. Specifically, they wanted to know how much overtime they would need to handle the transition and how much
smaller the number of workers could be afterward. The model developed demonstrated that the real transition constraint was parts usage, as this would spike when older parts were culled.
Air traffic safety improvement strategies
Air transportation safety is both very high and a matter of great concern. The agencies charged with improving this are also continually struggling with funding. One choice they must make is how much time
to devote to investigating accidents and incidents versus prospectively addressing issues. Though it seems desirable to get ahead of the next problem, it is actually much more efficient to investigate
things that have already happened. A small model, and even simple logic, easily demonstrates that to serve the future, scarce resources should be devoted to the past.
When launching a new product into a crowded but differentiated marketplace companies expect to take over existing market share as well as reach new customers. In a model used to capture the effects of
different pricing decisions and marketing activities it was observed that a product launch might in fact increase the sales of competitive products. This result, though surprising, was recognized as a
logical consequence of marketing efforts that made people more aware that products were available to serve specific needs. The company was able to use this insight, which was actually realized after launch,
to better understand their role in the marketplace.
Effectiveness of Agile software development techniques
The Agile software development process is designed to improve cost and delivery of projects in the face of uncertain customer requirements, but is it effective? A systems dynamics model was developed
comparing Agile methods to the traditional waterfall model and Agile was shown to perform better when customer requirements are unknown, but not consistently otherwise. In particular, the practices of
nightly builds, frequent releases, and avoiding complicated designs until proven needed showed the greatest impact in reducing the schedule and cost.
Does a subscription model lead to more stable revenues?
A small business was evaluating whether to switch from charging for new product updates and features to charging an annual subscription fee that includes all new updates and features, with the goal of
stabilizing the revenue stream. The system dynamics model showed that under very specific circumstances, an annual subscription stabilized revenues. However, if the conditions for these circumstances
were not met, the subscription model could further destabilize, and most likely decrease, the revenue stream. Detailed analysis indicated a number of places to improve data collection and reporting,
and also highlighted the importance of the subscription price relative to the cost of renewing a lapsed subscription.
Utility conservation programs in the Pacific Northwest
The nation’s electric utilities were facing serious problems with planning and financing in the 1970s and early 1980s. A surprising remedy for many of the problems was to purposely slow the growth in
electricity demand through utility conservation programs. Models were developed for the Bonneville Power Administration (BPA) during the 1980s-90s that helped BPA lead the way in the development of
major conservation programs. System dynamics delivered valuable insights over the course of many years.
Boom and bust in construction and the California electricity crisis
California introduced major changes in electricity markets in the late 1990s with the goal of encouraging a more competitive market for electricity. The new markets opened in 1998, and the state was
experiencing sky-rocketing wholesale prices and chronic brownouts/blackouts within just two years. Major distribution companies went bankrupt and many private generating companies went bankrupt thereafter.
System dynamics models were developed for the California Energy Commission to demonstrate the underlying reasons for the “Boom & Bust” pattern in construction of energy infrastructure and for the
vulnerability to sky-rocketing rates. The modeling results were used in California’s assessment of a new course to follow in the wake of the crisis.