MCG4308 Tutorial #3 - January 31, 2012

Question 1
You are given a simple mass-spring-damper system for design considerations: Given System Parameters and Constraints Mass: 2 kg % m % 3 kg Stiffness: k R 200 N/m Desired frequency range 8.16 rad/s n % 10 rad/s Initial conditions: x0 = 0 v0 % 300 mm/s Problem: design this sytem (choose a c) so that the response is always 25 mm. I.e. choose c so that x t % 25 mm.

restart : choice of k and m Let's look at the allowable values of k and m to see what the restriction on natural frequencies means to the choices of masses and springs. First, we need to make sure we choose the mass so that with the chosen value of n, the value of k lies within the acceptable range. 2 k = m $ n 0 With the minimum and maximum allowable values of mass for the minimum frequency n = 8.16 rad/s k = 2 $ 8.16 k = 3 $ 8.16
2 2

= k = 133.1712 N/m = k = 199.7568 N/m

With the minimum and maximum allowable values of mass for the maximum frequency n = 10 rad/s. k = 2 $ 10 k = 3 $ 10
2 2

= k = 200 N/m = k = 300 N/m

Conclusion: if we choose n = 8.16 rad/s, then m must be 3kg. If we choose n = 10 rad/s, then m can be anywhere in the 2-3kg range and then k can be in the 200-300 N/m range. To keep things simple, we want to try the design with those 'end' values. Desired frequency range 8.16 rad/s n % 10 rad/s tells us that we must design c such that the system is an underdamped harmonic oscillator. System response

Known system response for an underdamped harmonic oscillator is given by x t = A $ exp K$ n $ t $ sin d $ t C : A1 = v0 C $ n $ x0 d :

A2 = x0 : A= A2 CA2 : 1 2 A2 = arctan : A1 For this problem, we are given x0 = 0. v0 v0 Therefore, A = , =0 0 x t = $ exp K$ n $ t $ sin d $ t : d d 1 K then x t = n $ v0 1 K
2 2

amd since d = n $

$ exp K$ n $ t $ sin d $ t

We want to find the "worst-case" scenario (i.e the biggest possible amplitude) and this will happen when the coefficient in front of the exponential and sine is as large as possible. v0 In other words, worst case possible is as big as possible and given the constraint of v0 % 300 2 n $ 1 K mm/s, then the worst case scenario is for v0 to be big 0choose v0 = 300 mm/s = 0.3 m/s. Thi is done because usually initial conditions are not under control of a designer which is why we choose the worst case scenario. The other constraint is 8.16 rad/s n % 10 rad/s, which is somewhat under our control. The worst case scenario is n as small as possible, which means n = 8.16 rad/s since it makes the amplitude as large as possible. We can design with this to start. We expect the response to look something like

We expect the solution to oscillate and die. So the biggest (maximum) displacement will happen at that first peak. We can find this peak with the help of a bit of calculus.

x t := A $ exp K$ n $ t $ sin d $ t t/A e

sin d t

(1.1)

To find the peak, differentiate x t and set it equal to zero: diff x t , t = 0 K n e A

sin d t CA e

K t n

cos d t d = 0

(1.2)

divide through by the exponential function: (1.2) simplify A $ exp K $ n $ t K n sin d t Ccos d t d = 0 Solve for the time that makes equation (1.3) true tmax = solve (1.3), t arctan tmax = d n d
2

(1.3)

(1.4)

To check if this is a maximum or minimum, plug this value into the second derivative and simplify simplify subs t = rhs (1.4) , d = n $
K

1 K , diff x t , t $ 2
arctan 1 K 2 1 K 2

, power

Ae

1 K n 1 K
2 2

(1.5)

1C

Equation (1.5) is nasty looking, but as messy as it is, is is pretty clearly going to be a negative number for an underdamped system where 0< <1, which tells us that the value we got in equation (1.4) will be a maximum. You can also verify your work by plotting your answers at the end. Ok, so substitute what we know so far into x t and call that xmax : v0 2 xmax := subs t = rhs (1.4) , A = , d = n $ 1 K , v0 = 0.3, n = 8.16, x t 2 n $ 1 K
arctan K 1 K 2 1 K 2

0.03676470588 e 1C 1 K
2

(1.6)

For reasons apparent only to the designers of Maple, I'm going to replace the with r : xmax2 := subs = r, xmax

r arctan K

1 K r2 r 1 K r2

0.03676470588 e r 1C 1 Kr r2
2

(1.7)

If you are wondering why I changed from using to using r instead, it's because a bit of trail and error and frustration led to the fact that Maple would NOT solve equation (1.6) for because apparently has some kind of weird built-in definition and no, it didn't give me any sort of proper complaint -- it just refused to solve. So a bit of error so trail and error was involved. Things you learn. :-) Now we want that maximum amplitude (equation(1.7)) to be less than the maximum allowable value of 0.025 (use fsolve for a numerical solver and tell Maple what range you want the solution to be in, i.e. 0<r<1): = fsolve xmax2 = 0.025, r, 0 ..1 = 0.2889774788 (1.8) We can find the damping coefficients from c = 2 $ m $ n $ . Since we choose n = 8.16, we argued that it implied that m must be 3kg for that choice, so using that we get c = 2 $ 3 $ 8.16 $ rhs (1.8) = c = 14.14833736 kg/s Would no-damping ( = 0 solution have worked? FYI, if you were temped to try the no-damping solution with = 0, then the displacement would have been given by v0 x t = $ sin n $ t n v0 and clearly the maximum possible value of displacement would be which with a given maximum initial n velocity of 300mm/s and a maximum displacement of 0.025, would give 0.300 = 0.025 n 0.300 = 0.025 n which implies the natural frequency must be n = solve (1.9), n n = 12. and this is clearly not within the allowable range of 8.16 rad/s n % 10 rad/s. Repeat the analysis with n = 10 rad/s instead, which would also imply 2 kg % m % 3 kg As mentioned earlier, the worst case scenario is for v0 = 300 mm/s and usually initial conditions are not under control of a designer which is why we chose the worst case scenario. The system natural frequency is still somewhat within our control so let's redo the design but now with n = 10 rad/s. xmax3 := subs t = rhs (1.4) , A = n $ v0 1 K
2

(1.9)

(1.10)

, d = n $

1 K , v0 = 0.3, n = 10, x t

 arctan K

1 K 2 1 K 2

0.03000000000 e 1C 1 K
2

(1.11)

xmax4 := subs = r, xmax3

0.03000000000 e r 1 Kr2 1C r2

(1.12)

= fsolve xmax4 = 0.025, r, 0 ..1 = 0.1251549568 We can find the damping coefficients from c = 2 $ m $ n $ . Using the smallest possible value of mass gives (m= 2 kg which would imply k = 200 N/m) c = 2 $ 2 $ 10 $ rhs (1.13) = c = 5.006198272 kg/s Using the largest possible possible value of mass gives (m= 3 kg which would imply k = 300 N/m) c = 2 $ 3 $ 10 $ rhs (1.13) = c = 7.509297408 kg/s So, by designing with both end conditions of the allowable frequency range and the end possiblities on the mass range, we know that we can choose a damper somewhere in the range of 5.006198272 % c % 14.14833736 kg/s and we can find corresponding masses and springs to make the design work within the required constraints. The actual choice might be dictated by available choices of masses, springs and dampers but now we know the ranges that work. Check that the potential designs work First Design Des1 := subs A = n $ plot Des1, t = 0 ..5 v0 1 K
2

(1.13)

, d = n $

1 K , v0 = 0.3, n = 8.16, Zeta = 0.2889774788, x t

0.03840314150 e

sin 7.811860907 t

(1.14)

0.025 0.020 0.015 0.010 0.005 0 K 0.005 1 2 t 3 4 5

Check the maximum value of the first design maximize Des1, t = 0 ..5 0.02499999998 Second Design Des2 := subs A = n $ plot Des2, t = 0 ..4 v0 1 K
2

(1.15)

, d = n $

1 K , v0 = 0.3, n = 10, Zeta = 0.1251549568, x t

0.03023775321 e

sin 9.921372066 t

(1.16)

0.02 0.01 0 K 0.01

2 t

Find maximum displacement in the second design maximize Des2, t = 0 ..4 0.02500000000 We can compare the responses of the two potential designs plot Des1, Des2 , t = 0 ..4

(1.17)

0.02 0.01 0 K 0.01

2 t

Clearly both work but the responses are different due to the different levels of damping and different frequencies. simplify subs t = rhs (1.4) , d = n $
K

1 K , diff x t , t $ 2
arctan 1 K 2 1 K 2

, power

Ae

1 K n 1 K
2 2

(1.18)

1C

Question 2
Damping coefficients are difficult to measure and required dynamic, not static, tests. The logarithmic decrement provides a good way to estimate the damping ratio , from which c can be found. For an underdamped system, the logarithimic decrement is defined as the natural logarithm of the ratio of two successive amplitudes: x t = ln x t CT 2 where T = is the period of the underdamped system. In other words, you measure the position at some time, d and then measure the position again one period (T) later. Show that if you can measure then can be found from = 2 2 4 C

For an underdamped system x t := A $ e

$ sin d $ t C t/A e
K t n

sin d t C

(2.1)

= ln

x t x t CT

 = ln e

K t n

sin d t C sin d t CT C = d T = 2 so that

K t CT n

(2.2)

T is a a period of the system so that d $ T =

d $ 2 $ d

sin d t CT C = sin d$ t C C2$ = sin omegad$ t C (2.2). Same for the e

and the two sine terms will cancel out in

, they will cancel out in numerator and denominator. So equation (2.2) becomes = ln e
$ $T n

= $ n $ T

Substitute for T = subs T =

2$ , and d = n$ d
2

1K , then we get

2$ , d = n $ d

1K , = $ n$ T = 2 1 K
2

(2.3)

Solve equation (2.3) for in terms of (take only the positive root ) to get: solve (2.3), 4 C
2 2

,K

4 C
2 2

(2.4)

The math gives us two roots which are the same thing. Remember is a positive number, so we take only the positive root: = 2 2 4 C

Question 3
How is the previous formula modified if you make the measurements not one period but rather but n cycles apart?

suppose the peak values are denoted B1, B2, etc and note that

ln

B1 Bn C 1 B1 B2

= ln

B1 B2

B2 B3

B3 B4

$$$$ B3 B4

Bn Bn C 1 C...

= ln

Cln Bn Bn C 1

B2 B3

Cln

Cln

= CCC... = n $ So for any two points n cycles apart: =

1 x t $ ln n x t Cn $ T

Question 4
Measurement of the free response of a certain system of mass m = 450 kg shows that after 5 cycles, the amplitude of the displacement is 10% of the first amplitude. Also, the time for these 5 cycles to occur was measured to be 20 seconds. Estimate the system's damping c and stiffness k.

restart : From data n = 5 n =5 x t x t C5 $ T = x t 0.10 $ x t x t x t C5 T = 10.00000000 1 x t $ ln n x t Cn $ T (4.2) (4.1)

Logarthmic decrement, then from question 2 is given by = = evalf 1 $ ln 10 5

= 0.4605170186 Find the damping ratio from the logarithmic decrement evalf subs (4.3), = 2 2 4 C = 0.07309748496 Period is given by T = d = n$ 1K
2 2

(4.3)

(4.4)

2$ 2$ 20 seconds 1 = 4 seconds = so d = = d = 5 cycles 4 2 d

d = n subs (4.4), d = , (4.5) 2

1 K

(4.5)

1 = 0.9973248005 n 2 n = solve (4.6), n n = 1.575009792 m := 450 : n := rhs (4.7) : Stiffness is found from

(4.6)

(4.7)

k := m $ n = 1116.295130

N/m

Damping constant is found from c = 2 $ $ m $ k c = 2$rhs (4.4) $ m $ k = c = 103.6163291 Ns/m

Question 5
A boy riding a bicycle can be modeled as a spring-mass-damper system with an equivalent weight, stiffness, and damping constant of 800 N, 50000 N/m, and 1000 N-s/m, respectively. The differential setting of the concrete blocks on the road cause the level surface to decrease suddenly as indicated in the figure. If the speed of the bicycle is 5 m/s (18 km/hr), determine the displacement of the boy in the vertical direction. Assume that the bicycle is free of vertical vibration before encountering the step change in the vertical displacement.