Systems Model Series: System Dynamics (SD)

System Dynamics, Feedback Loops

Adapted from the article originally written by Yeni Mulyono

for the Cabrera's Cornell University, System Thinking in Public Affairs Course


System Dynamics (SD) is an approach to problem-solving which developed by Jay Forrester (1960). The system is a collection of elements that regularly interact over time to form a unified whole. Dynamics refers to change over time. System dynamic is a methodology used to understand how systems change over time. After that, SD also was popularized by Peter Senge (1990) in The Fifth Discipline: The Art & Practice of The Learning Organization. He focused on how to solve the group problems by using system thinking method in order to convert companies into learning organization. He also maps out the structure of the system using system dynamic (feedback loop).

We can use System Dynamics (SD) to make some analysis of the complex system. It can be employed for qualitative and qualitative modeling.


A. SD Qualitative (Feedback Loops)

Feedback loops is central of SD that allow nonlinear behavior. It has a function to capture the relationship between interactive subsystems. Feedback occurs when outputs of a system are routed back as inputs as part of a chain of cause and effect that forms a circuit loop.

Feedback Loops

Figure 1: Reinforcing and balancing feedback loops

Source: Cabrera, D. & Cabrera, L. (2015) Systems Thinking Made Simple (STMS), 2nd Edition, p147

The figure shows two feedback loops have interaction. A reinforcing loop called “word of mouth” shows that more adopters mean a higher adoption rate, which in turn means more adopters. The other feedback loop is a balancing loop because when potential adopters increases, the adoption rate climbs, but as the adoption rate climb, potential adopters decline. By using feedback, the behavior of a system can be altered to be fit with the application (can make the system stable, responsive, or constant).


B. SD Quantitative (Modelling)

System modeling technique was based on two building blocks:

  • These can be compared to reservoirs where the flow can be accumulated
  • Flows are comparable to a faucet that fills these reservoirs or a drain that drains them.

Modeling is used to conduct an empirical test of a hypothesis relating to the structure responsible for the performance overtime observed in selected variables. Under quantitative perspective, SD uses numerical data to calculate parameter values, characterize system behavior, and compare with model output.

Table 1: A Comparison between Qualitative and Quantitative perspectives in the modeling process

Modeling process

Qualitative SD

Quantitative SD

Objective of modeling

Understand the feedback structure of the system.

Test a hypothesis about the structure driving the reference mode of the variable under study.


Text data obtained through facilitated face-to-face meetings, interviews, or the interpretation of causal mechanisms in reports and from theories.

The structure of the model is developed using similar inputs employed in qualitative SD. Additional data for the model can take three sources: judgment from experts or clients, numerical datasets for parameters, and facilitation processes for nonlinear functions.


The modeling process implies the construction of causal loop diagrams to represent individual and/or group- level interpretations of causal links.

Then, equations are formulated and parameters populated using the inputs. Testing of structure and outputs are performed to confirm the hypothesis that the structure is able to replicate the behavior observed.


There are three main outputs: learning about the structure of the system, changes in participants’ perspectives, and agreement on future policies.

There are three outputs. First, time-series showing performance over time of relevant variables. Second, policies are tested quantitatively to improve the reference mode. Third, learning about dynamic behavior of the system.

Source: Kunch, Martin, “System Dynamics : A Soft and Hard Approach to Modelling”, Warwick Business School, p4.


The difference between SD and other quantitative methods is the efforts to identify all causal mechanisms responsible for the behavior of a model made by the modeler (Kunch 2017).

Even though SD was popular, there are some pros and cons regarding the SD method. The pros explained that SD technique strongly emphasized the dynamics of studying the whole system to understand its behavior, SD also provided system thinking with a firm methodology tested by computers. But, the cons of SD described that SD derived from an electrical engineering perspective that divided a system into sub-components, the approach is flawed when the whole is more than its parts. Furthermore, the assumption in SD that described the individual behavior, eliminated the effects of luck, noise, and randomness in a complex system. The modeling technique should be used only for complex situations whose outcomes lead to great impact.



  1. Cabrera, D. & Cabrera, L. (2015) Systems Thinking Made Simple (STMS), 2nd Edition
  2. Forrester, J. W. (1997). Industrial dynamics. Journal of the Operational Research Society48(10), 1037-1041.
  3. Kunc, M. (2017, December). System dynamics: a soft and hard approach to modeling. In 2017 Winter Simulation Conference (WSC)(pp. 597-606). IEEE.
  4. Senge, P. M. (2006). The fifth discipline: The art and practice of the learning organization. Broadway Business.
  5. Sweetser, A. (1999, July). A comparison of system dynamics (SD) and discrete event simulation (DES). In 17th International Conference of the System Dynamics Society(pp. 20-23).