SpaceManMat
Space Nut
- Joined
- Dec 20, 2013
- Messages
- 702
- Reaction score
- 86
Does anyone know why open rocket optimization keeps intermitately returning zero altitudes?
Open Rocket Otimization issue
- Thread starter SpaceManMat
- Start date
Help Support The Rocketry Forum:
- Joined
- Sep 5, 2009
- Messages
- 3,534
- Reaction score
- 1,660
Optimization routines are tricky and can quickly go haywire, depending on the starting point and range of interest. It is best to start small. I would ignore the zero points and just look at the top points. Seems like increasing your override mass beyond 75g does not help, but it says nothing about less than 75g.
- Joined
- Aug 18, 2011
- Messages
- 1,459
- Reaction score
- 57
Just looking at the graph and thinking about the math behind what OR is doing, I would say the issue is math.
The only thing you need to worry about would be the top points of the graph.
The only thing you need to worry about would be the top points of the graph.
SpaceManMat
Space Nut
- Joined
- Dec 20, 2013
- Messages
- 702
- Reaction score
- 86
I find the results a little disturbing not entirely sure if I want to trust the results. Will certainly compare with other simulators. I find it often treats the ranges I setup as mere suggestions and refuses to show me the other half of the results, sometimes there are issues with the weights being set (such as stability) but it does not indicate that. One optimization I did set the stability to .5 even though I set the limit at 1, like I said a mere suggestion. Anyway not trying to sound too grumpy would just like reassurance that the "good" values I am getting are actually meaningful.
- Joined
- Sep 5, 2009
- Messages
- 3,534
- Reaction score
- 1,660
The plot clearly shows altitude vs. electronics mass.
SpaceManMat
Space Nut
- Joined
- Dec 20, 2013
- Messages
- 702
- Reaction score
- 86
I suspect you will find this all over the place when using the optimizer. Reading a bit on the optimization of OR it is pretty complex. To me the error appears to be either the optimizer choosing not to sim some values and then mistakes this for a zero result or there is something not being intialised between simulations that can sometimes result in a failed simulation.
The other glitch I sometime see is when the current designed value is outside the range you want to optimize for. When this occurs OR can show a completely blank graph (zero simulations). To get around this I first modify the design so that what I want to optimize is in the middle of the range I want to use.
A nice to have change would be that the configuration defaults to the first one, it would be nice if it defaulted to the motor you had loaded. I keep forgetting to select the motor, then I optimise and realize I forgot to change the motor. Due the issue with the current value (above) I cannot get a result out of the optimizer. I then need to close the optimizer (if I had hit apply then I need to undo that), only then can I give it another shot (would not be much of an issue if it wasn't for the other glitch).
The other glitch I sometime see is when the current designed value is outside the range you want to optimize for. When this occurs OR can show a completely blank graph (zero simulations). To get around this I first modify the design so that what I want to optimize is in the middle of the range I want to use.
A nice to have change would be that the configuration defaults to the first one, it would be nice if it defaulted to the motor you had loaded. I keep forgetting to select the motor, then I optimise and realize I forgot to change the motor. Due the issue with the current value (above) I cannot get a result out of the optimizer. I then need to close the optimizer (if I had hit apply then I need to undo that), only then can I give it another shot (would not be much of an issue if it wasn't for the other glitch).
- Joined
- Sep 5, 2009
- Messages
- 3,534
- Reaction score
- 1,660
The only way to know is to re-run and your simulations with the proposed optimized values. Given the troubles you have with the optimizer routine, you might be better off just manually iterating through your parameters of interest, anyway.
bobkrech
Well-Known Member
- Joined
- Jan 20, 2009
- Messages
- 8,352
- Reaction score
- 55
You gave no numerical information on how you conducted your optimizations but the above reply has the information I was going to suggest.
I find the current input window for changing the optimization parameter inputs confusing, and that's probably the source of the strange plot. The OR optimization seems more complex than it needs to be IMO. A simple variation of +/- total mass is what you need to consider for most cases and in the simplest case leave it to the designer to decide where the mass reductions or additions will happen.
Bob
wcravens
Well-Known Member
- Joined
- Jun 29, 2015
- Messages
- 129
- Reaction score
- 0
I've not jumped on this earlier because I didn't have much time. But I could look now.
Could you attach an OR file and instructions on how to demonstrate your concerns? I'll be glad to look into it after that. I've been wanting an excuse to dive into the optimization internals for a while.
Wes
Could you attach an OR file and instructions on how to demonstrate your concerns? I'll be glad to look into it after that. I've been wanting an excuse to dive into the optimization internals for a while.
Wes
SpaceManMat
Space Nut
- Joined
- Dec 20, 2013
- Messages
- 702
- Reaction score
- 86
I would rather keep that particular sim to myself for now. I should be able to come up with another example I'm sure.
SpaceManMat
Space Nut
- Joined
- Dec 20, 2013
- Messages
- 702
- Reaction score
- 86
As request, attached zip file has the OR sim. There are 2 print screen in there that show the values you need to set.
View attachment Optimization Errors.zip
View attachment Optimization Errors.zip
wcravens
Well-Known Member
- Joined
- Jun 29, 2015
- Messages
- 129
- Reaction score
- 0
That's perfect. In these situations it is in fact much better to boil everything down to the absolute bare bones minimum that still demonstrates the issue.
Cool. I'll have a poke and see if I can report back something useful.
Happy Rocketing!
SpaceManMat
Space Nut
- Joined
- Dec 20, 2013
- Messages
- 702
- Reaction score
- 86
thanks wcravens
- Joined
- Sep 5, 2009
- Messages
- 3,534
- Reaction score
- 1,660
The optimization function looks very powerful at first blush. Almost any rocket design parameter can be selected, which seems cool. However, as bobkrech noted, trying to do a simple mass optimization is difficult. Looks like you need to create a phony design file of zero mass and then add a mass object to select as the parameter for iteration. If the input screen would allow an override mass, then this would be a lot easier. Rocksim's optimal mass routine is much more simple and logical.
For debugging, I suggest unchecking the stability constraints and keep it simple.
How about environmental parameters? Launch rod angle is a good parameter with a sweet spot for altitude or range, yet is not available in the input list.
Last edited:
bobkrech
Well-Known Member
- Joined
- Jan 20, 2009
- Messages
- 8,352
- Reaction score
- 55
With regard to the optimum mass, you also limited the stability margin to 1. The optimum mass with the stability margin option = 1 is 3.82 oz. so there is no solution below this value. Remove the stability margin requirement and the optimization will be the minimum mass, but the resulting rocket design might be unstable.
Bob
