Discuss:Science, Engineering, Knowledge:Systems Analysis
From da Vinci Concept
Systems Analysis
carrie - Tue Feb 10 13:22:01 2004
I was just wondering how many of you have heard of or worked with Systems Analysis, also known as Systems Dynamics. I was introduced to this last year at my master's program and we used it to model and simulate different problems, mostly related to sustainable development, but it can be used for most anything. It is both a way of thinking and a problem solving tool. There are various programs, such as Stella, that can be used to make working models.
I was so excited to learn about this because it is the one problem solving method I have heard of where you can really include all the variables, even time and human behaviour (if you can successfully measure that!).
So if anyone has worked with this I would be curious to know. I will send on more information if anyone is interested. There is an institution in Austria (IIASA) that deals with if you want to look into it, and more I am sure. As I said, I will try to sit down and write more details about this another time!
--carrie Tue Feb 10 13:22:01 2004
henrik - Sat Feb 14 9:18:52 2004
I have never heard of this before, but if you have more info I would be curious to know.
--henrik Sat Feb 14 9:18:52 2004
sunny - Tue Feb 17 20:32:07 2004
I think that I have heard of something similar that I read about in a science fiction novel years ago concerning general approaches to system design and a sort of system level engineering degree that covered all of the bases. The work that I was reading was describing how the prevalence of specialists were drowning out the discipline of Systems Engineering. The 'Systems Dynamics' that you are describing seems a similar sort of science.
A couple of Google searches brought a plethora of relevant results:
Looking further into the IIASA site:
And particularly:
- Managing Problems of Postmodernity: Some Heuristics for Evaluation, by Darek M. Eriksson, Interim Report, © 1998, IIASA
Upon reading just the first few pages it becomes apparent that I am not the only one thinking about systemic aproaches to problem solving. The IIASA foundation also appears to be starting from the ground (metaphysic->epistemologic) up. This foundation and the system of systems aproaches it describes are exactly the type of methods I think are relevant to a more general aproach to problem solving. Upon consideration, sustainable development seems the obvious candidate for this type of systemic thinking. Within the cited paper's abstract I agree with the caveat that the religious motives of much of societal systemic thinking is a cause for an aparent impotence in these methods. I disagree with the implication within the final statement of the abstract, that motives for systemic thinking for sustainable development need be religious in nature. But I definitely agree that "religious ground-motives . . . [make] any prospect of human advancement pessimistic." I will provide a more detailed discussion upon full reading of the document.
Carrie:
Thanks for the great information!
Did you have a textbook for the course, or provided any pertinent reference material?
What methods, in particular, did you employ in your research? In application, do these include heuristic and statistical methods?
Others:
Has anyone been introduced to similar methods? Is there a System Dynamics psuedonym for more specialized fields or hard sciences?
[ Edited Wed Feb 18 2004, 12:05AM ]
--sunny Tue Feb 17 20:32:07 2004
carrie - Fri Feb 20 4:35:13 2004
More about Systems Dynamics/Analysis...
I donít know if any of you have ever seen the movie Mind Walk, but systems thinking uses much of this idea - considering problems in the whole and within context. It is typical to dissect problems and focus on the problem point...but it is like solving a stomach ache with some Pepto-Bismol when you really have a food allergy. As such, academic disciplines have been divided and the interconnectedness of problems across these is often overlooked. Using systems analysis/dynamics is one way to help bridge this. It requires a shift in the brain to see problems not as linear cause and effects, but rather as a system of feedbacks, where one action can reinforce or counterbalance another. And it is a way of seeing problems over time and the changes that occur or develop, rather than in the ìmomentî.
There are all levels of modelling, but it is best to start simple, from the beginning. It is not difficult; you can begin by drawing out your problem on a piece of paper and identifying all of the possible related factors within the limits that you set (your model boundaries). The example below is taken from a lecture. Then you can decide if the factors have a positive correlation (if unsafe practices go up, the number of accidents go up, or they both go down) or a negative correlation (if traffic volume goes up, speed goes down, if volume goes down, speed goes up). This is a causal loop diagram.
PLEASE SEE BELOW!
This takes much trial and error, especially for a complex problem, and on the professional level, it requires the consultation of a range of experts that know much more about, for example, traffic, driving psychology, accident statistics, etc. For a good working model, you really have to have ALL of the information about EVERYTHING within your model boundaries!
You can then take this Causal Loop Diagram you created, and put it into a modelling program. I have used Stella, but there are others such as Dynamo and PowerSim. Here is an example, again taken from a lecture...
POSTED BELOW
You see the simple causal loop diagram of a population, with balancing factor of resources. The pink and blue is the Stella model, and the graph shows the behaviour of the system over time. Once you have made a working model, then you can begin playing with the variables, testing the model itself and the whole system. In this case of a population, you could try changing the resource availability, or pretend you are in China and put in a control on the birth rates, or even throw in a disease.
During my program I used causal loops to model a variety of things, such as a community waste problem (involving policies, economics, limited space, methane recovery, etc) aging populations (with their working class shift, social security, etc.) and so on. Systems thinking has been completely ingrained in us by now! We did not have a text book but I do have some of the lecture presentations that I can try to upload on this site. One particularly good one is on system archetypes ñ the system pattern that you will see repeated in many different situations.
If you would like to learn more, the Systems Dynamics Society http://www.albany.edu/cpr/sds/index.html has some online courses you can take yourself through.
A relevant book is the ìThe Fifth Disciplineî by Peter Senge which talks about the five disciplines: systems thinking, personal mastery, mental models, shared vision, and team learning.
There is even a conference this summer July 25 ñ 29 at Keble College, Oxford, England. I would really like to go to this, if anyone is interested maybe I can meet you there!
[ Edited Mon Feb 23 2004, 06:27AM ]
--carrie Fri Feb 20 4:35:13 2004
carrie - Mon Feb 23 3:24:30 2004
This is the first image I was referring to - this a CLd done in Vensim.
Source: Systems Analysis Lecture - Mats Svensson, Harald Svedrup, Hˆder Haraldsson. LUMES, Lunds Universitetet vÂrtermin 2003.
[ Edited Thu Feb 26 2004, 10:32AM ]
--carrie Mon Feb 23 3:24:30 2004
carrie - Mon Feb 23 3:26:28 2004
This is the secong image - here there is a CLD, and the pink and blue model done in STELLA, and a graph result from STELLA showing the model over time.
Source: Systems Analysis Lecture - Mats Svensson, Harald Svedrup, Hˆder Haraldsson. LUMES, Lunds Universitetet vÂrtermin 2003.
[ Edited Thu Feb 26 2004, 10:32AM ]
--carrie Mon Feb 23 3:26:28 2004
sunny - Tue Feb 24 22:05:41 2004
These diagrams are very reminiscent of my analog control systems theory courses. These types of feedback models are very interesting in engineering disciplines as they allow for the use of 'black-box' type modelling where the input and output are known, but not all of the particular internal states are known. This is partcularly useful when a state is not observeable/measureable. It is also the case that because certain states within a control system are not observable, these same states are not controllable, meaning that since you cannot measure them you cannot use them directly within any feedback loops for control.
I am going to do some reading regarding the Systems Dynamics material and am very interested in learning the current applications.
-----
Regarding posting lecture notes, please see: Forum Guidelines: Submitting and uploading copyrighted material
[ Edited Wed Feb 25 2004, 03:03AM ]
--sunny Tue Feb 24 22:05:41 2004
Truc_Ha - Wed Feb 25 19:11:06 2004
Fascinating. The diagrams also remind me of classwork, though mine refers to systems that interact in the field of medicines. One method I use in problem solving is to mentally build such diagrams and take them through various states: eg. the various states of a disease, or waxing and waning of a drug's effects, or even just physiologic shifts throughout the day. I am interested in making a serious study of this for both pedagogic purposes and to clarify and streamline my own mental processes.
I am wondering: in the context of a professional school (medical school) I wonder what kind of module could be developed to give students an understanding of the mechanism of problems solving in general. We do emphasize the commonality of certain disease processes, for instance, cancers go through the progression hyperplasia-metaplasia-neoplasia. What do you think?
Also, I tried a google search "systems+dynamics+medicine" and didn't get very useful links. Any suggestions on getting more information on systems dynamics in medicine and medical education?
[ Edited Wed Feb 25 2004, 10:11PM ]
--Truc_Ha Wed Feb 25 19:11:06 2004
carrie - Thu Feb 26 7:30:00 2004
I know that the Santa Fe Institute www.santafe.edu is looking at Cancer as a complex adaptive system. It is part of this summer school program. And the IIASA and Santa Fe study robust systems, which could be related to health as well. Other than that I have not heard of any particular medical applications of this.
--carrie Thu Feb 26 7:30:00 2004
cbettles - Mon Mar 15 11:19:32 2004
This is really interesting. I don't have much experience with it and I will have read the Heuristics paper Sonny mentioned. Here ares some of my first impressions:
In the area of sustainable development, it is certainly better than the current method of looking at things in isolation. The setup of the EPA is a good example. Having a seperate office for things like Air pollution, Water pollution, and Solid Wate (Air, Water, Earth...pretty sophisticed) tends to ignore how they interact to form complex environmental problems. Now that we've done the easy stuff (shutting down large centralized pollutors) we really need tools like this to tackle the more complex problems.
Like any other model, however, the end result is largely predicated on what you put in the model and what questions you are asking.
[ Edited Mon Mar 15 2004, 02:20PM ]
--cbettles Mon Mar 15 11:19:32 2004
