Sunteți pe pagina 1din 11

Accuracy of modelling projectiles using classical mechanics

Kyle Byrne
Abstract This report will be an eort into nding the accuracy available in modelling the paths and behaviour of prejectiles using classical mechanics. I hope to achieve this by carrying out a set of projectile experiments, lming and then from this extracting the parabola formed by the path of the projectile. I will then compare this to the predicted parabola as found by classical mechanics, described below. By comparing these paths I will have a visual representation of the aect air resistance and other external forces have on the projectile.Through the values I record I will also be able to extract numerical values for characteristics of the ight which I can compare to my real world ights as well. The motivation behind this subject is to nd out just how good of an approximation we can nd when leaving out values such as air resistance.

Background Physics

To model a projectile with initial velocity = u red from an angle to the horizontal we must rst split it into its horizontal and vertical components. Doing so can give uvertical = u sin anduhorizontal = u cos . From these we can 2 simply substitute into s = ut + at 2 to nd the horizontal and vertical distances. svertical = y = u sin t gt2 2
x u cos

shorizontal = x = u cos t as ahorizontal is 0 Rearranging the horizontal component we can nd that t = tuting this into the vertical we can nd g x x u cos u cos 2 x2 g u2 cos 2 y = x tan 2 gx2 y = x tan 2 2u cos2 gx2 sec2 y = x tan 2u2 y = u sin
2

and substi-

This result will provide the basis for the visual comparison side of this report by allowing us to predict what path the projectile should take. We can derive the time of ight and horizontal range by using the equation 2 y = ut + 1 2 at using the vertical component of velocity we get 1 y = u sin gt2 2 And so From this we can see that t = 0 or t = 2u sin g 1 t(u sin gt) = 0 2

this time value is known as our time of ight , it is the time taken to fall back to an initial height after projection. To nd the x co-ordinate at this time and hence the range the particle travels we can this time value into the equation S = uhorizontal t we then nd that Srange = u cos t u cos 2u sin Srange = g 2u2 sin cos Srange = g u2 sin 2 Srange = as 2 sin cos = sin 2 g

With these values I will be able to both visually and numerically compare the predicted ight with the real ight. Values for the real ight will be found by tracking the particle through its ight at correct scale, this will give us actual x , y ,t and S values for any time with our recorded velocity and angle.

Preliminary trials and research

Through initial testing it became clear that there was a number of problems with the experiment , specically ; Method of projection Distance measuring Velocity measuring

2.1

Method of Projection

The rst problem I came across was how I would actually re the projectile, I shortlisted 4 dierent methods. Elastic bands Sling (e.g. trebuchet) Air powered Spring By assessing the characteristics I was looking for I could rule out 3 of these methods. Elastic bands would be too unreliable, not providing consistent repeatable power. The sling method would add rotation to the projectile, meaning it would interact with the air dierently and so any progression into calculating the eect of air resistance would prove far more dicult. The air powered method whilst probably being the most repeatable would have the problem of being probably too powerful, being dicult to carry out the low power testings I hoped to do, also this method would be the hardest to get hold of the apparatus for. So this leaves Spring powered, I managed to get hold of a spring powered cannon which provided suitable, repeatable and changeable power output, and also the

added bonus of easy aiming and angle measurement.

(1)

2.2

Distance Measuring

The premise of my experiment was about comparing visually and so for this it became quite apparent quite soon that I would need high speed video to be able to actually extract any sort of data from the footage. The problem with this however is that as you increase FPS you decrease in resolution and so this meant that my original plan of having metre rulers in shot for measuring would no longer work, as they werent readable. The solution I came up with was using a striped tape measure with width of 4cm not for measuring but instead just to use as a scale, this would all be handled by the software I was using to track the projectiles, a piece of software made by Cabrillo college which would not only allow me to track the position of the projectile for each frame but also let me set a scale for the scene so any measuring I needed to do could be done on the computer against this scale. The software would also provide me with the parabola I needed from the video in the form of a polynomial which I could then plot against my calculated parabola to compare.

2.3

Velocity Measuring

With the cannon I had got hold of I had one of the values I needed sorted, the angle. This just left me with nding the velocity of the projectile, for which I had 2 choices. Either I could do it by assessing the footage or I could do it with hardware. Through software I would be able to use coordinate geometry to nd the distance the projectile travelled through a time period and from this calculate the velocity, the downside to this would be my scale, as discussed previously the footage from the slow motion camera wasnt of great resolution meaning any sort of accurate measuring was out of the window (mainly due to blur on the projectile so not knowing its exact location) which would leave this method of nding distance with great uncertainty. Which is why I opted for the alternative, building a piece of hardware carry out the measuring.

What I came up with was a light gate style addition to the cannon which would mount through the rails on the top and bottom as visible in the above picture. The barrel extension would contain 2 infra red LED traps connected to a micro controller , as the rst gate was broken a value of the time since the start of the program running would be saved and then the same thing would happen for the second gate. This gave the time taken between the two for which I knew the distance meaning I could calculate the velocity between these 2 points. I could then get instant read o on my laptop of the results. Below is the schematic for the device , the program I wrote to run it with a description of how it worked, and also a copy of how I received the results from the device.

Risk assessment

With the obvious dangers apparent when using projectiles in an experiment I took the following precautions to ensure my safety and safety of those around me. I wore eye protection at all time and asked those in close proximity to my experiment to do so as well A barrier was set up around the area I was working in to stop people wandering across the ight path unaware People around me were alerted when I would be ring the projectile A back panel was put up to stop the projectile travelling outside of the ring area

4
4.1

Journal
-Lesson 1

Lesson 1 was primarily spent getting the equipment working ,nding a suitable location for where I would be carrying out the experiment and how Id be set up. 6

It was after seeing the materials available to me in this lesson that I progressed the aim of my project from investigating solely air resistance based on shapes of projectiles to a more broader focus due to limited numbers of projectiles available to me.

4.2

-Lesson 2,3

Lessons 2 and 3 were used for preliminary testing making sure everyone wold work as expected and trying to reduce error as much as possible. This is spoken about more in the Preliminary trials section.

4.3

-Lesson 4

Lesson 4 was spent putting into place the solutions I had came up with in lesson 3 and verifying they would all work. This included the new scale and testing out my infrared barrier.

4.4

-Lessons 5-9

Lessons 5-9 were spent gathering the data , doing 5 repeats for each of the 3 power levels for varying angles.

5
5.1

Analysis
Visual analysis

My primary method of analysis is through visual comparison, comparing the graphs from the video and from calculation. As mentioned earlier I am using the tracking software from made by Cabrillo college to track the video. This gives me a polynomial for the best t curve of the parabola, it also allows me to nd x and y components for any time value, nd acceleration at any point etc whilst also working under scale set up using my markers in the real scene. With over 60 graphs I couldnt include them all in the report and so the ones being shown are just examples for each power and angle ,typical of what the others showed. By comparing the various angles it is apparent that with the smaller angles the approximation hugs tighter the actual curve than those at 60 or 50 degrees, I believe this is due to the fact that with less of a vertical velocity component the smaller angles are less aected by initial air resistance on re, when the velocity is greatest ,due to the fact that the projectile is spending more time at its lowest speed when travelling at a higher angle.. And then air resistance in the horizontal direction will aect each projectile equally independent of their angle. When comparing the dierent power values it is visible that the highest power level has far more accuracy than the lower power values, and so a velocity travelling at a higher velocity is a better approximation. This could be due to the fact that with the high velocities it has a higher kinetic energy, giving it a better chance to overcome the resisting force of air resistance. One thing we would expect to see with the real life experiments which isnt so apparent in these results is the shape change we would predict with air 7

resistance. What we would expect is to see a parabola not travelling as far and with a lower maximum point. Whilst we are seeing often a shorter distance travelled it seems usually that in fact the real world projectiles have a higher maximum point than those which have been predicted. Figure 1: A predicted model of air resistance on a projectile

5.2

Time of ight

Using the formula t = 2u sin I derived a value for the time of the ight of the g projectile. I then calculated the % error between this calculated result and the real result, found by analysing video frame by frame and then multiplying by 1 120 where 120 is the frame rate of the camera. To nd a value for those where the projectile left the frame before falling back to the same height I tracked the projectile and then using the best t parabola found the root of the generated equation.

The rst graph in the time analysis section of graphs shows these errors. What you can see is a quite consistent error for each angle and power of roughly 10% this implies that some correction could be applied to the equation which would make it more accurate, the problem is that whilst the graph doesnt show which direction the error is in by evaluating the raw data table it can be seen that whilst mostly an overestimate there was also time where the calculated value would be smaller than the predicted value and so in these cases a correction could not be applied across each experiment. To understand why most of the predicted values were an overestimate we must input air resistance into our model, this would act as a resisting force on the projectile (opposite to its direction of ight) decreasing its vertical acceleration (as force and acceleration are proportional) meaning a lower greatest height is 8

achieved, due to the fact the projectile reaches 0 velocity quicker. This lower greatest height would mean less distance travelled overall rather than the height predicted by ignoring air resistance, less distance would mean less time taken to get back to the height of projection.

5.3

Range of ight
2

2 To calculate the horizontal range of ight I used the formula Srange = u sin 2g derived earlier. Again I calculated the % error for each experiment and plotted them on a graph. This graph again shows a net constant average error however it shows much greater spread than with the time values. The average error is almost double that of the time of ight at 20%. The recorded values were found using similar methods as before. By looking at the calculated distance against actual distance for each power graphs it seems that in this example it is not the prediction that is at fault. The graph shows some anomalous results in the 40 degree experiment where when you would expect this to have an , on average , greater range than that of the 30 degree experiment it fell short, but then the 50 and 60 degree experiments followed the predicted trend line. Implying that my setup was open to systematic error aecting the results. With these calculations it can be seen that the calculated graphs again usually overestimate the distance. This would again be down to air resistance which would result in a shorter parabola.

Evaluation

This experiment was riddled with systematic error. The most major of which was the barrel extension I made, whilst it did a good job at measuring velocity and agreed in that sense with every other calculation I did to check its validity the added length it gave to the barrel caused problems of deection. Especially at the lower power levels the projectile would hit the end of the barrel, ring at an angle dierent to that which the launcher was set at. Whilst sometimes this could be noted and a repeat could be done other times it was less noticeable especially when it happened at higher velocities. This could be in part responsible for the extra lift visible on the real life projectiles. Another theory I have for this unaccounted for lift is the projectile itself, the projectile was a small wooden ball, wood is prone to having an irregular structure, with small knots throughout it, these irregularities could have put spin on the ball, in a vertical axis along its direction of travel, this spin could easily result in the phenomenon of the Magnus eect. Where turbulence is left behind the projectile causing a lift due to pressure dierence, similar to the way plane wings work. To counteract this it may have in fact been useful to use a projectile method such as air projection, where a tighter barrel would be needed as I could use riing like in a gun barrel to add spin in a vertical plan perpendicular to the direction of motion which may have stabilised the projectile reducing the eect of any irregularities in its weight. The fact also that some of my projectiles were travelling o frame before completing their parabola meant that I was having to extrapolate from what I could see, increasing uncertainty. This led to real problems in the 60circ test where I didnt have enough tracking points to really nd any sort of curve of the 9

parabola before it left the top of the screen, and so no data could be found for these ights. One of the subjects I would like to explore if I was to redo the experiment is modelling the air resistance on the projectile also and seeing how accurately this could be modelled using the formula Fair = 1 ACu2 2

where is the density of air, C is the drag coecient depending on the shape of the particle and A is the cross sectional area of the projectile. I would also make an eort to nd a more regular and repeatable method of ring which also added spin to the projectile.

6.1

Uncertainties

The theoretical uncertainty of my light gate setup was 0.000004 seconds (for its timing value), due to this being the cycle rate of the micro controller I was using however taking into account the spread of the IR LEDs I was using and so the uncertainty of when each beam would be broken I took the uncertainty in the velocity value to be 0.005ms1 . I believe this to be the lowest uncertainty I could achieve with the equipment to hand. The major uncertainty was in the angle value, to measure the angle I used the built in protractor and plum line on the launcher, whilst this can be quite an accurate way to measure it is prone to parallax and also with each re there would be a slight movement of the barrel, taking this into account I gave the angle an uncertainty of 3 . This could have been improved by using a more sturdy base for my launcher or by nding some other method of measuring the angle, less likely to shift and without parallax. When nding the range of the projectile I gave the distance an uncertainty of 0.02m , this was due to the fact that my scale was made from 0.04m strips, With my timing I managed to reduce the error by using a slow motion camera, shooting at 120FPS rather than the usual 30 or 25 this managed to give me an 1 uncertainty of 120 th of a second as I was able to go through frame by frame to nd when it crossed the axis I had set up at ring height in the tracking software.

Conclusion

To conclude my report the experiment has shown that the the approximation of classical mechanics whilst not perfect does give a reasonably good model for the real life motion of projectiles. Especially for smaller angles of ring and high initial velocities. However I do believe this conclusion lacks in some strength due to the sporadic nature of the data I received and often high uncertainties. Even with this taken into account though I believe that for most circumstances they provide a suitable model ,unless very high precision is needed.

References
Mechanics 2 for OCR textbook ISBN-10:0521549019 10

http://www.britannica.com/EBchecked/topic/357684/Magnus-eect http://www.wired.com/wiredscience/2012/01/projectile-motion-primer-forrst-robotics/

11

S-ar putea să vă placă și