Stella Helps Eskom Keep the Lights on in South Africa
Most of us only think about electricity when we don’t have it. We’re annoyed when a power outage keeps us from binge watching
our favorite show or reading the new bestseller. We’re worried when inefficient power generation and delivery block school students
from technology, present challenges to medical professionals and their patients, or throttle manufacturing productivity. The staff
at Eskom thinks about electricity every minute of every day; about sustainably producing it, about efficiently distributing it,
and about keeping power generation plants and the grid up and running.
Eskom generates, transmits, and distributes almost all the electricity – about 95% - used in South Africa and about 45% of the
electricity used on the African continent. Their market dominance positions Eskom as a strategic contributor to South Africa’s
economic development and growth and mandates that the company provide electricity in an efficient and sustainable manner.
“We need to be able to make long-term decisions in a globally turbulent environment that is characterized by constrained planetary
resources,” says Nalini Sooknanan Pillay, Head of the System Dynamics Centre of Expertise, in the Eskom Research, Testing, and Development
business unit. “Many of our conventionally established business models are threatened by current conditions and that requires us
to move away from a linear problem-solving approach toward one that allows us to build in causality and feedback loops and
identify unintended consequences.”
In 2010, Eskom began using System Dynamics to support the Scenario Planning initiative driven by Eskom’s Corporate Strategy and
Risk Division. Executive Managers recognized it as a beneficial approach to strategy, analysis, and decision making and the
company widened its use. Pillay’s team now uses Stella from isee systems to build strategic and operational simulations which
include time delays, causality, and feedback loops. Those features support both quantitative and qualitative analyses.
“Stella is easy to use,” says Pillay. “It allows us to build project interfaces that facilitate stakeholder engagement because
they are integrated with structural modeling components. That makes a huge difference when we’re explaining scenarios and patterns
of customer and operational behavior to executives and other stakeholders. The advanced Stella modelling skills developed by our
cross-disciplinary team over the years has ensured the delivery of high quality models which are on par with the best in the world.”
The projects taken on by the System Dynamics Center of Expertise (CoE) are initiated by other Eskom divisions engaged in long-term
planning. Typically, the divisional teams need to factor multiple criteria into the decisions they make. Projects are proposed
to an Investment Committee that approves a budget for model development.
A structured system dynamics modelling approach, entrenched by the CoE, guides all Eskom projects. Problems are carefully
articulated and assumptions, boundary conditions, and reference modes documented. A dynamic hypothesis is formulated and a
model is built. System architecture maps together with causal loop diagrams are developed to highlight system boundaries,
key driving forces, and causal elements and make it easier for the CoE team to engage non-technical stakeholders in
simulation findings and insights.
The CoE staff delivers a wide-range of models that support critical strategic decisions. Examples of some of these models that
look across the electricity market and operational landscape include:
- a Scenario Simulator that considers driving forces affecting four possible futures for the electricity industry,
- a Management Flight Simulator that describes the relationships between the Customer Service Index, distribution network
performance, and other factors (e.g. trouble response time, demand, average interruption duration and frequency)
that will impact customer satisfaction over the long term, and
- a Plant Performance Simulator that links a range of variables that impact plant availability.
Other models address more operational issues like the project interdependency simulator, the hydro pumped storage dynamics,
and the demineralized water production planning tool.
“Using the system dynamics models we build, Eskom is able to understand system complexities and are far more likely able to
make better operational and long-term strategic decisions,” says Pillay. “The simulators allow strategic dialogue among
decision makers and increase understanding of how short-term decisions impact the long-term health of the organization.”
The Management Flight Simulator displays Customer Value Perception across regions and market sectors
Improvement scenarios are adjusted to see the impact on service interruptions over time
For example, a simulator that looked at factors impacting customer satisfaction surprised senior management by pointing
out that short-term improvements in the distribution network don’t have an immediate impact on customer satisfaction.
In fact, it takes 3 to 5 years for improvements in the network to yield an increase in customer satisfaction.
The simulator looks at network improvements and customer satisfaction over the long term (2007 to 2035). The System Average
Interruption Duration Index (SAIDI) is used to identify the percentage change in network improvement. Feeder site improvement
scenarios and a variety of fault time measures (e.g. repair times, dispatch and travel times, etc.) are factored into that index.
Eskom gauges customer satisfaction levels by market sector using a periodically administered MaxiCare survey.
“It was intuitive that there was a relationship between the duration of service interruptions and customer satisfaction but
the actual time delay was unexpected,” says Pillay. Understanding the relationship between distribution improvements and
customer satisfaction set appropriate expectations and provided critical guidance for network planning.
Another simulator focuses on the supply of demineralized water that is essential for generating thermal power. It runs over
170 hours, i.e., just over one week, and considers the demineralization trains and process that ensures water is available
for generation. Users input values to describe the number of online vessels, their flowrates and initial and maximum throughputs,
brine wash frequency, regeneration durations, the station’s demineralized water consumption, extra water consumed by blow downs,
maintenance, and water required to light up (start) units.
“The Stella tool we developed assists demineralized water planning,” says Pillay. “It helps managers ensure that there is
enough water available for light ups and make better planned outage decisions.”
Demineralized water production planners consider the relationship between lightup intervals and tank levels
Pillay points out that Eskom’s dominance in the South African market gives it a huge advantage but the insights delivered by
Stella models and simulators are essential to formulating sustainable strategy. “We’re in a unique position to be able to
oversee the electricity generation-to-end-user value chain but that’s a complex system,” says Pillay. “The use of contemporary
approaches like system dynamics is critical. We’re able to merge Eskom’s empirical history with scenario-based simulations of
future trends, conduct sensitivity analyses, understand the behavioral characteristics of our systems, and formulate tracking
metrics. And that leads to much better operational and long-term strategic decision making.”
TexPREP and Stella Engage Young STEM Students in Systems Thinking
The more we understand the complexity of current world problems, the harder we cross our fingers that high school and college students
are preparing to solve them. Inspiring more students to study and enter careers in science, technology, engineering, and math (STEM)
is one way to make sure the next generation is equipped for the world they’re inheriting. As important, STEM students need to practice
taking a big picture, Systems Thinking approach to understanding problems and finding the best solutions.
“Systems Thinking is a core skill that needs to be taught as early as students are able to understand it,” says Ben Jurewicz, Adjunct Professor,
Saint Mary’s University. That opinion is the product of Jurewicz’s education which includes graduate work at MIT and long tenure as a systems
engineer. “I’ve worked in aerospace; I was in the Apollo program and was VP of Engineering at Sperry Corporation, and, as a consultant,
applied system dynamics to IT, financial services, electronics, and other issues. Through all those experiences, I encountered very few
people who understand whole systems. It’s a real deficiency in our educational system.”
Jurewicz is building the Systems Thinking capacity of tomorrow’s engineers and scientists through the graduate level courses he teaches
at Saint Mary’s University and TexPREP, a San Antonio, Texas based summer education program designed for middle and high school STEM
students. “Eight years ago, TexPREP added a fourth year, Prep IV, to the program in order to teach students how to solve complex problems
using a systems approach,” says Jurewicz. “We use Stella as a tool for understanding systems and encouraging the kind of ‘What if’
thinking that students will need to become good scientists and engineers.”
“We learned that we can’t teach Systems Thinking in the abstract,” says Jurewicz. “We have to teach from the ground up using an example
that they care about.” TexPREP integrates systems theory and water science classes. The water science class focuses on the physics and
hydraulics of rainfall and consequent stream flow and uses the local Salado Creek Watershed as an example system. “First, students learn
how heavy rainfalls in the watershed area can lead to significant water flow in the stream and cause serious flooding,” says Jurewicz.
“Understanding how that process happens is the challenge for Systems Thinking.”
It has taken TexPREP several years to develop a teaching process that successfully teaches Systems Thinking concepts while engaging students.
Today, the program takes a three step approach that begins with a conceptual definition of a system.
“They learn that a system definition depends on an individual’s perspective and that it’s important to understand how the parts of a
system contribute to the whole,” says Jurewicz. Examples including cars, electronic devices, and human organizations are used to prompt
discussion of how a system fits in any environment, its internal elements, and internal and external relationships that influence behavior
and outcomes. “We talk about how systems have a purpose and exhibit behavior that changes over time.
We challenge students to predict behavior under given conditions and they quickly realize that isn’t as easy as it seems.”
Second, students learn to isolate elements of the system before being introduced to causes and effects. “Once students understand that
causes and effects can move in the same or different directions we talk about feedback loops and how they are driven by causes, effects,
and time delays between them,” says Jurewicz. “We build Causal Loop Diagrams to see qualitative internal relationships and create a
foundation for modeling quantitative relationships.”
Third, a Stella model that includes stocks and flows teaches students how to quantify the impact of feedback loops and delays. “We start
with a very simple model; water flowing in and out of a bathtub or population levels with births as inflows and deaths as outflows,
and run model simulations to explore behavior of stocks and flows over time,” says Jurewicz. “These simple models demonstrate that
outcomes are a consequence of systems structural relationships. They prepare students to take on the bigger challenge of modeling
the behavior of output as a consequence of a rainfall event in the watershed they’ve been learning about.”
Students begin their modeling project with a Causal Loop Diagram that defines it as an element in a much larger system.
Causal Loop Diagram created by TexPREP students
Their study of watershed behavior allows students to model the physics of water transport through the Salado Creek Watershed system.
Their work begins with a very simple inflow and outflow model that demonstrates water accumulation. They exercise the model to
illustrate non-intuitive relationships among included variables.
Simple Stella model of the Salado Creek Watershed
Students then add elements to develop a more realistic, complex model that is calibrated using real data from USGS Salado Creek Station 13.
Students see that their model computes peak stream flows that correspond closely with actual measured flows.
The final Salado Creek Watershed Stella model
When their calibrated model is complete, students are challenged to explore “what if” scenarios. “Problems are rarely static,” says Jurewicz.
“They are present when something causes something else to happen. The model pushes students that aren’t natural ‘what if’ thinkers to take
a more dynamic approach to problem solving.”
Students are asked, for example, “What if more dams were built to increase the system’s water retention capacity?” Again, they use real data
to show the impact of damming that adds 50% more retention and a delayed slow release. (Curve 1 is the flow curve with the added dam.
Curve 2 is the original flow.)
The impact of water retention and slow release
Students see that, with additional damming and slow release, stream flow drops below normal levels after about 160 hours and are challenged
to explain why that happens. They also explore the impact of rainfall at different rates and over different periods of time on flood levels
and how watershed urbanization influences runoff and flooding.
While TexPREP participants are all considered high-achieving students (they have to carry a B or above average), they come to Systems Thinking
with a wide range of natural inclination or interest. Even so, pre- and post-testing indicates that Prep IV instruction leads to real learning.
“We test students on core concepts and see consistent positive results,” says Jurewicz.
That’s good news for a world filled with complex problems in need of well-trained engineers and scientists. “When people don’t understand the
system that underlies a problem, they tend to take the first solution that comes to mind and that’s often the wrong solution,” says Jurewicz.
“Our students are learning that the broad view is as important as detailed, narrow views. Finding the best solution will require them to
understand a whole system and all of its parts in order to identify and evaluate options.”
This article uses material from Prototyping Systems Thinking Curriculum Development for Pre-College Students, Ben R. Jurewicz which was
published in the Proceedings of the Third International Conference on Complex Systems Design and Management, 2012.
Systems in Focus
Systems are all around us. Using Systems Thinking, we can better understand them by building models that can be simulated. These videos
show common systems overlaid with the underlying dynamics that make them behave as they do. By looking beneath the surface, we are better
able to bring these systems into focus. An example of a simple bottle manufacturing process is our most recent System in Focus.
The modern assembly line is one of the greatest and now most common manufacturing process concepts in history. It was created to save
time and money, and to increase the overall quality of the output. By mechanically moving parts in a line through a factory to different
work stations, a product can systematically be built up piece by piece, while decreasing time spent moving parts. This process was
perfected and used most famously by Henry Ford in the 1920s
In this simple example, we have modeled the basics of a bottling plant. Before we can model the process however, we need to consider
the supplies required to create the final product, in this case, water. A bottling plant requires a constant supply of purified water.
It is usually run through three sets of filters that remove smaller and smaller particles before being sterilized.
The bottling side starts with bottles that have either been shipped in (glass) or are blown into the proper shape from small stubs of
plastic. Glass bottles are placed on a disk-shaped hopper while plastic bottles are often dumped into an open box hopper. The hopper
feeds a machine known as an unscrambler, which puts the bottles onto a conveyor belt in single file. From there, the bottles are
quickly rinsed to ensure cleanliness, filled with water, and capped.
The bottles are then inspected for stray particles inside and some bottles are rejected. Since filling made the bottles wet, they
are blown dry before labels are affixed to them. At this point, any labels that need to be printed onto the bottle are added.
Finally, bottles are grouped and packed into containers. In the case of water, the package gets shrink-wrapped.
The package of water is now ready to ship!
To see other Systems in Focus visit the main page Here
From Imagination to Simulation
A scraping, kind of shuffling noise woke me from a dreamless sleep. I didn’t know where I was and I didn’t recognize the sound. My smartphone told me it was 2 am and by its
illumination, I saw the stacked hay bales surrounding me. In a flash, it all came back to me. The mad race from Chicago to New York to London to Paris and finally to this small,
remote farm in the French countryside, the one place I thought I could escape them. It wasn’t true, after all, that they were only in the cities. They’ve taken over the entire
world with their scraping, shuffling gait and their dead, dead eyes. There must only be a few of us left now… How much longer can we hold out?
This is just one possible outcome to the…zombie apocalypse! Welcome to our new modeling series, From Imagination to Simulation, which investigates out-of-the-ordinary
problems and examines creative “What If?” scenarios. This series showcases the versatility of Systems Thinking while demonstrating how fun it can be along the way!
Join us this Halloween on this creative adventure as we discover the possible outcomes that could arise from a zombie outbreak. We hope that you enjoy (and learn from) this new experience!
Stella Professional Keeps Getting Better
With the latest update, Stella Professional continues to improve upon it's user experience! From the new Import Panel to creative new ways to add variables to graphs and tables, users
will be pleased with the new features and attention to even the smallest of details.
Still haven't tried Stella Professional or want to learn more? Join us for a free webinar on Tuesday, October 27th from 12:00 – 1:00 EDT in which you will get a full “show and tell” demonstration from one
of our leading developers, with time for questions and answers at the end.
Throughout the presentation, you will see the similarities to isee systems’ original STELLA and iThink software, as well as many of the new features and functionality that you’ve
been requesting over the years.
Using the latest development tools available, our software engineers have transformed the systems thinking software that you know and love, to create the next generation of modeling tools.
Built from the ground-up, Stella Professional has a sleek, modern look and feel, designed to help you intuitively find what you need, while maintaining a familiarity to previous versions.
Our developers have carefully crafted a multitude of new features, including Stella Live, to allow you to immediately and more precisely analyze the data in your models. For users of
STELLA/iThink, Stella Professional simultaneously maintains compatibility with the models you’ve been developing over the years and effortlessly brings your model to the future.
Come see our most powerful tool yet, Stella Professional, we are excited to share it with you!
Click Here to register.
On the Road!
This autumn will be a busy season for isee co-president Karim Chichakly! First, Karim will be attending the 3rd Brazilian Systems Dynamic Symposium
and Workshop in São Paulo, Brazil from October 16-17th! Karim will be a guest speaker at the symposium and discuss why projects consistently fail and
how systems thinking can help. The following day he, along with fellow systems thinker Andrew Ford, will instruct a full day workshop for attendees.
This workshop will focus on engaging stakeholders and clients in system dynamic models! To learn more on this Symposium and workshop
Directly following the Brazilian Symposium, Karim will hop over to Columbia for the 13th annual Congreso Latinoamericano de Dinamica de Sistemas (CLADS)
in Cartagena de Indias, Columbia from October 21st – 23rd. This will be our second time attending this conference and we are excited to once again showcase
our software and meet fellow practitioners of the Latin chapter of the Systems Dynamic Society. Karim will give a talk at this conference as well as conduct
introduction and intermediate workshops for attendees. For more information on CLADS visit
Advanced Modeling Series
Are you comfortable with the basics of dynamic modeling? Would you like to advance to a higher level? Do you have specific applications you’d like to model,
but are not quite sure where to start? Our new Advanced Modeling series offers courses centered on the application of dynamic modeling to specific topic areas.
Common problems and approaches are presented, along with sample models to get you started!
Our first course in the series, Modeling Projects, highlights some of the competing forces that make successful project management a challenge, as well as
some ways to counteract these forces. Modeling Projects is already available in prerecorded form Here.
The second course, Strategic Business Development, explores strategic management issues and the challenge of turning strategic objectives into winning
operational objectives. Strategic Business Development starts Wednesday, November 4. Register Here!
isee systems now offers 2- and 6- month subscriptions to our STELLA/iThink 10.1 and Stella Professional Software. This new packaging offers allows
users to choose the modeling software that fits their project, time frame, and budget. Each limited-duration subscription for a non-academic license
can be purchased and renewed in periods of either two or six months.
Click Here to Buy