need help now-- human modelling simulation

[farhana Syed]

hi, I am really in confused , studying about human modelling simulation in DES, and try to insert DETAIL human factor in DES...such as their skillness, behaviour, impact of fatigue, worksload, ....

BUT after a year i found that DES is not a right platform to captured human factors...

Please Give me your opinion..?..I NEED YOUR SUGGESTION OR EVEN YES/NO ANSWERS FOR THESE FOLLOWING QUESTION.....THANKS ALOTS


1. Human Modelling cannot be full captured by DES???

2. ARE Human Modelling Simulation is a CONTINUOS SIMULATION??

3. hAVE EVER HEARD ABOUT SYSTEM DYNAMICS??

4. SYTEM DYNAMIC Is a SUITable knowledge TO CAPTURED HUMAN BEHAVIOUR RATHER THAN DICRETE EVEN SIMULATION (DES)..IS THAT TRUE...??

5. DO YOU HAVE ANY IDEAS ABOUT HUMAN MODELLING IN SIMULATION...??


PLEASE REPLY...

[ewilliams]

Greetings.
In my opinion and experience, modeling of humans can be captured partly, but not fully, by DES. For example, DES can accommodate data or hypotheses such as "during the last two hours of the shift, the worker works 10% slower due to fatigue," or such as "worker #1 is more skilled than worker #2; either #1 or #2 can be assigned to a task but worker #1 will need 5% less time," or clients would rather deal with bank teller #1 than bank teller #2 because, although they are equally efficient, teller #1 is more friendly; therefore clients will go to teller #1 if both are free when the client enters the bank. DES cannot adequately, by itself, model situations such as the onset of carpal tunnel syndrome, rotator cuff injury, or hypothermia due to the cumulative effect of adverse working conditions (these are continuous processes within the worker's body), nor is it the best tool to model issues such as "if this workstation is implemented as designed, only 10% of males and 3% of females can undertake the reaching required without repetitive stress injury, but a redesign could raise these percentages to 80% of males and 60% of females." DES often can model interactions among humans (such as "which passport line will the next traveler approach or which pathway through a subway station will the next traveler take, given observable behavior such as where the last few travelers have walked"). To capture such interactions, agent-based discrete simulators (such as AnyLogic) are often more appropriate than process-based simulators (such as Arena). Most importantly, as you undertake the IE work you have briefly described, I urge you to eschew the thought "I must choose one analytical technique (e.g., DES, ergonomics, value-stream mapping, scheduling algorithms....) in favor of a holistic approach in which you have many analytical approaches in your mind's toolbox and many analytical software tools available on your computer or in your computer laboratory.

[kvasudevan]

Hi,

In addition to Mr.Williams's comments.... I would like to add that:
1. Humans can never be modeled fully. We have to use an abstraction dependent on the application. If you are more interested in the decision, then complex decision tools can be used to make the model logic. If you are interested in ergonomic studies through simulation, then you may need to model the graphic rich motion; inch-by-inch etc..

2. Classification of Human modeling is again dependent on the application. If we choose to model complex decisions at discrete points of time, then it needs a DES engine.... If we want to model the path of motion of an arm or torso etc, then we need to use differential equations that border on continuous simulations

3. System dynamics is 'the most abstract' form of simulation modeling. It works on levels and flow of entity groups. It gives little individuality to components. The questions it can answer include what percentage of the population will become engineers? It cannot answer the question - Will Mr.ABCD become a engineer?

4. System dynamics can be used to capture population behavior if that is what you are interested in. If on the other hand, you are interested in individual behavior or evolving crowd behavior etc, then agent based simulation is the right approach.

5. Please refer to http://www.informs-sim.org/wsc05papers/002.pdf for a useful intro into Agent based simulation.

In conclusion, ask yourself the question, What is most important to the application? Is it to model decisions/behavior, patterns/trends or intricate motion? Based on the answer, I would go for ABS, System Dynamic simulations, Pseudo Continuous simulations respectively.

Hope this helps..

Karthik Vasudevan, PMC

[farhana Syed]

Thanks a lot for the help, I am really appreciate it,

Based on both of you suggestion

I think now I know the main problem of my research, I am confuse in choosing the best method. maybe because of 2 different point of views from both academician and profesionals..

Actually my task is to give the result of the VARIABILITY impact of human in the operation.. And discrete even simulation (DES) has been chosen as a method. Even though is limited ability to capture human but I thinks it’s just enough for this study for 1 year.

I am also realized that there are a lot of human factors that will affect the productivity…for examples:

Fatigue, skills, experiences, schedule pressure-shift,emotion or physical factors and a lot more…

And each of these factors has a numbers of levels. It is getting more complicated because I have to assign all the variables one by one in the DES coding , and see the output on the production line.
There is no data or hypotheses given, and I have to do the direct observations in the production line. .

I hope the method that I’ve choose is suitable enough to finish this research.. did both of you agree??

Moreover, I thinks this study can be as detail as you wanted too, its can be ergonomic study, DES, or behavior(continuos)…..if you have any opinion or informations regarding this please inform me…because I am still learn about the human modeling in simulation.
It is such an interesting topic to discuss.
Thanks again

Regards,

[ewilliams]

Greetings; you are on a good track now. From your further descriptions, it seems clear to me that you will also want to consider the statistical aspects of your work very closely. As you mention, ambient conditions (temperature, humidity, noise) can all affect a person's speed of work, as can fatigue, emotion, and other things you mention. You may wish, as applicable to your situation, to examine factors such as the length of queue (e.g., does a supermarket checkout clerk work faster and spend less time in friendly banter when he or she can see that the line is long?). To help in this statistical analysis, you may wish to avail yourself of distribution fitters. As you may already be aware, there are specialized software tools which, given a set of data values, will examine them to decide whether an exponential, a gamma, a Weibull, a ..... (various distributions) represents a good statistical fit for the data. If you find such a fit, the mathematical properties of the fitted distribution can be used to make further inferences concerning the data. Further questions welcome.

[farhana Syed]

yes its really need detail statistical data..
as for the software what do u think the best DES software for modellling this kind of probs...

Automod?? maybe....??

regards

[kvasudevan]

Good to hear you are on the right track.

Regarding the software, what are your options? AutoMod is statistically powerful. But Arena for instance has a great input data analyzer/curve fitter that comes with it.

If you are looking to build a manufacturing system model with workers and then model worker cycle time based on worker charecterstics (using a combination of distributions and formule with dependent/independent variables) then AutoMod may be a suitable way to go, but note this can be done in prettymuch any simulation software. AutoMod gives you good code control through functions and processes if that is what you desire.

Remember though, the real power of AutoMod lies in its handling of Material Handling Systems. Unless you are planning to use the kinematics module in AutoMod to model human actions, you might not be justifying the use of AutoMod power.

- Karthik, PMC