Dear all,
Is there a probability function in AIMMS?
For example, I want to calculate:
where:
x: The capacity of power plant to be constructed (MW)
y: Installed capacity of power plant
F: force outage rate of power plant
Thank you very much….
Solved
Probability function in AIMMS
+4Best answer by Marcel Hunting
That depends how you use LOLP in your model. Note that a chance constraint in AIMMS is of the form:
You could vary epsilon and solve the robust optimization model multiple times, and find the "optimal” epsilon in this way. LOLP would the correspond to 1 - the "optimal” epsilon.
Practically, this could only work if the deterministic model can be solved (reasonable) fast, because the robust optimization model will be harder to solve, and you have to solve robust optimization model multiple times for different epsilons. Also, I do not understand the role of t; do you have a probability function for each t?
+4Hi
I guess x and y are variables, and F and D are parameters. One of the parameters F and D must be uncertain (or both) otherwise using a probability function does not make much sense.
Does LOLP have to be calculated as part of the optimization, or after the optimization? In the latter case you could generate multiple random values for F and/or D, assuming you know the distribution(s), and then check whether the constraint is satisfied for (x,y) or not. If you do this many times then you will get a good approximation for the probability.
If LOLP has to be calculated as part of the optimization then I see no way to do this in AIMMS. Note that AIMMS supports chance constraints as part of robust optimization but that is not the same.
+4Dear
Thank you for your reply.
You’re right, x is a variable of power plant capacity should be built to meet the demand D. y is the capacity of existing power plant. While F is the parameter of unavailability factor of power plant.
The LOLP in this problem is must be calculated as part of the optimization. So, based on your answer, I cannot use chance constraints in AIMMS.
Is there a way to solve this optimization model?
+4That depends how you use LOLP in your model. Note that a chance constraint in AIMMS is of the form:
You could vary epsilon and solve the robust optimization model multiple times, and find the "optimal” epsilon in this way. LOLP would the correspond to 1 - the "optimal” epsilon.
Practically, this could only work if the deterministic model can be solved (reasonable) fast, because the robust optimization model will be harder to solve, and you have to solve robust optimization model multiple times for different epsilons. Also, I do not understand the role of t; do you have a probability function for each t?
Sign up
Already have an account? Login
Please use your business or academic e-mail address to register
Login to the community
No account yet? Create an account
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.