5

Views:
 
Category: Entertainment
     
 

Presentation Description

Liked by Godwin Thiruchelvam

Comments

Presentation Transcript

Helicopter Flight Time:

Helicopter Flight Time In teams of 4/5 Experiment with the design of a helicopter to result in maximum flight time Money is only available for only 10 prototype helicopters What changes will you test to determine the best design? Time 60 Minutes

Helicopter Design:

Helicopter Design Target: Maximise flight time Y (seconds) Changes allowed: (Design Parameters) X’s (below) X 1 : Wing length 90mm 115mm X 2 : Wing Chord 55mm 65mm X 3 : Weight 1 clip 2 clips X 4 : Stiffness card paper X 5 : Body length 85mm 100mm X 6 : Body width 35mm 45mm X 7 : Wing joint tape no yes 110 or 130 mm 35 or 45 mm 85 or 100 mm 35 mm 90 or 115 mm Paper clip Standard Maximum

Helicopter:

Wing Joint Tape Helicopter

Strategy and Results:

Proto Paper No. Taped Body Body Wing Wing Flight Type Type Clips Wing Width Length Chord Length Test Result Strategy and Results 1 2 3 4 5 6 7 8 Confirm

You have 60 minutes:

You have 60 minutes You only have 10 attempts How do you choose which experiments to run? Go !

Who had the best time ?:

Who had the best time ? Did The winner only use 10 prototype helicopters? How do they do this in the real world? The helicopters cost real money to build They cost real time to build You can’t just keep building new helicopters to find the best one

“One Factor at a Time”:

“One Factor at a Time” Set baseline combination of conditions Change one factor If better, keep the change and try next factor If results not improved, revert and try next factor

“One Factor at a Time” Strategy:

“One Factor at a Time” Strategy Experiment Paper type No. Clips Tape Body width Body Length Wing Chord Wing Length 1. Standard – – – – – – – 2. Paper Trial + – – – – – – 3. Clip Trial – + – – – – – 4. Taped Wing Trial – – + – – – – 5. Wide Body Trial – – – + – – – 6. long Body Trial – – – – + – – 7. Wide Chord Trial – – – – – + – 8. Long Wing Trial – – – – – – +

“Stick With a Winner” Strategy:

“Stick With a Winner” Strategy Experiment Result 1. Standard – – – – – – – 2.0 2. Paper Trial + – – – – – – 2.4 3. Clip Trial + + – – – – – 2.2 4. Taped Wing Trial + – + – – – – 2.2 5. Wide Body Trial + – – + – – – 2.4 6. Long Body Trial + – – + + – – 2.6 7. Wide Chord Trial + – – + + + – 2.7 8. Long Wing Trial + – – + + – + 3.0 Paper type No. Clips Tape Body Width Body Length Wing Chord Wing Length

“One Factor at a Time”:

“One Factor at a Time” The results from changing one factor at a time could be plotted as the changes in the output Y against the change in a particular factor X The “One Factor at a Time” approach can help to find the best design But do the settings of the other Xs affect the amount of change in Y due to change in the selected factor? Y Paper type X 1 Y Wing Chord X 3 Y Body length X 2 Y Wing Span X 4

Will these approaches always work?:

Will these approaches always work? Interactions between factors can and often exist. Indeed, interactions can have a greater effect on the output than changes in individual factors One factor at a time experiments cannot identify interactions between the Xs If the optimum performance depends on interactions one factor at a time experiments will not find the optimum Results can depend on the order of the experiments! So “One Factor at a Time” or “Stick with a Winner” do not always work !

How do you find the best design ?:

How do you find the best design ? “ One Factor at a Time” and “Stick with a Winner” approaches are good They can reduce the number of experiments you need to perform ie reduces cost However they do not always work If the factors are inter-related Result can depend no the order you carry out the tests

“Test them All”:

“Test them All” The best approach may be to test all possible combinations How many prototypes would this need for our helicopter? If you assume that each factor can only take on 2 values ie Body width 35mm or 45mm (not 40mm) We need 128 prototypes to test every combination That is way too many – way to expensive

How do you find the best design? :

How do you find the best design? How do you find the best design? every time? With minimum number of prototypes? Lowest cost? Look up “Design of Experiments” on the web – you might find a way

authorStream Live Help