Damping, Friction, and Energy Dissipation in the Kinematic Chain
Your body is not a frictionless machine. Every joint has viscous resistance. Every muscle contracts against internal friction. Every tendon absorbs and releases energy with a bit of loss. When you swing a golf club, only a fraction of the energy your muscles generate actually reaches the clubhead. The rest vanishes into heat, joint compression, tissue deformation, and the subtle resistance of bodily motion.
This chapter asks three fundamental questions: Where does this energy go? How does damping at one joint propagate through the entire kinematic chain? And crucially — is damping a bug to overcome, or a feature your nervous system harnesses for control and stability?
The answer is: it depends on how much damping you have and at which joints. Too little, and you can’t control the swing. Too much, and you waste precious energy. The art of an efficient golf swing is tuning this damping landscape.
Adding Damping to the Equations of Motion
Recall from Chapter 9 that the unforced kinematic chain is governed by: \[ \massmat(\bm{q})\ddot{\bm{q}} + \cormat(\bm{q},\dot{\bm{q}})\dot{\bm{q}} + \gravvec(\bm{q}) = \bm{B}(\bm{q})\control + \bm{J}_c^T(\bm{q})\constraintforce \]
This is the drift vector field — the natural motion of the body under gravity, Coriolis effects, and muscle forces. To include damping, we simply add a damping term:
\[ \boxed{\massmat(\bm{q})\ddot{\bm{q}} + \cormat(\bm{q},\dot{\bm{q}})\dot{\bm{q}} + \bm{D}(\bm{q})\dot{\bm{q}} + \gravvec(\bm{q}) = \bm{B}(\bm{q})\control + \bm{J}_c^T(\bm{q})\constraintforce} \] where \(\bm{D}(\bm{q})\) is the damping matrix, typically diagonal or nearly diagonal, with entries \(D_{ii} \geq 0\).
The damping matrix \(\bm{D}(\bm{q})\) has entries that represent the damping torque at joint \(i\) per unit angular velocity. For a joint with pure viscous damping from synovial fluid and muscle, \(\bm{D}\) is constant (independent of \(\bm{q}\)). But in reality, \(\bm{D}\) depends on configuration: - When a joint is flexed deeply, more muscle is in parallel with that joint, increasing damping. - Tendon routing changes with posture, affecting effective damping. - Soft tissue compression increases with extreme angles.
So \(\bm{D}(\bm{q})\) is a function of joint angles, even if its variation is modest.
Types of Damping Models
The simplest model is linear (viscous) damping: \[ \bm{f}_d = -\bm{D} \dot{\bm{q}}, \quad D_{ii} = \text{constant} \]
In reality, damping is often nonlinear. Common models include:
Coulomb (kinetic) friction: \(f_d = -\mu_k |f_n| \,\text{sign}(\dot{q}_i)\). Force is constant, direction-dependent. Unphysical for continuous motion but useful for large-scale analysis.
Quadratic damping: \(f_d = -\rho A C_d |\dot{q}_i| \dot{q}_i\), proportional to velocity squared. Common in aerodynamics and high-speed fluid flow.
Power-law damping: \(f_d = -c |\dot{q}_i|^n \,\text{sign}(\dot{q}_i)\) for \(n \in (0, 2)\). Many biological tissues exhibit this mixed behavior.
Stribeck damping: A more realistic model with a velocity-dependent friction coefficient: \[ f_d = -(\mu_s - (\mu_s - \mu_k)e^{-|\dot{q}_i|/v_s}) \,\text{sign}(\dot{q}_i) \cdot f_n \] where \(\mu_s\) is static friction, \(\mu_k\) kinetic, and \(v_s\) a characteristic velocity. This captures the smooth transition from stiction to sliding.
For most of this chapter, we will use linear viscous damping because: 1. It’s analytically tractable. 2. Experimental evidence suggests it’s a reasonable approximation for velocities in the middle range (not too slow, not supersonic). 3. When velocities are high (downswing), the linear term dominates; when slow (address), the system is nearly quasi-static anyway.
Rewriting in Control-Affine Form
In the control-affine setting, recall that the ZTCF family is: \[ \dot{\mathbf{x}} = f_0(\mathbf{x}) + \sum_{i=1}^m g_i(\mathbf{x}) u_i = f_0(\mathbf{x}) + G(\mathbf{x})\control \]
The undamped drift field is \(f_0(\mathbf{x})\). When we add damping, we modify the drift: \[ \dot{\mathbf{x}} = f_0(\mathbf{x}) - \bm{D}_{\text{eff}}(\mathbf{x})\dot{\bm{q}} + G(\mathbf{x})\control \]
The “effective damping” \(\bm{D}_{\text{eff}}\) includes not just \(\bm{D}(\bm{q})\) but also the implicit damping from the Coriolis terms, because changing velocity changes the Coriolis forces. This is a subtle point: damping is not simply additive; it modifies the entire vector field structure.
How Damping Propagates Through a Chain
This is the key insight. Damping at one joint does not affect only that joint. It propagates through the entire chain via the mass matrix coupling, Coriolis coupling, and constraint coupling.
Direct Effect: Local Damping
The most obvious effect is local. Damping at joint \(i\) directly opposes angular velocity at that joint: \[ -D_{ii} \dot{q}_i \] This reduces the rate of change of \(\dot{q}_i\) by an amount proportional to \(D_{ii}\). No surprise here.
Coupling Through the Mass Matrix
Here is where it gets interesting. Rearranging the damped equation of motion: \[ \ddot{\bm{q}} = \massmat^{-1} \left[ \bm{B}\control - \cormat\dot{\bm{q}} - \bm{D}\dot{\bm{q}} - \gravvec \right] \]
Because \(\massmat\) is not diagonal, the terms inside the brackets get “mixed” when inverted. Let’s illustrate with a simple two-joint chain: shoulder (joint 1) and elbow (joint 2).
The mass matrix for a two-link arm is approximately: \[ \massmat = \begin{pmatrix} m_1 L_1^2 + m_2(L_1^2 + L_2^2 + 2L_1 L_2 \cos q_2) & m_2(L_2^2 + L_1 L_2 \cos q_2) \\ m_2(L_2^2 + L_1 L_2 \cos q_2) & m_2 L_2^2 \end{pmatrix} \]
The off-diagonal terms \(M_{12} = M_{21}\) are nonzero and depend on the elbow angle \(q_2\). When we invert this matrix to solve for \(\ddot{\bm{q}}\), we get: \[ \ddot{\bm{q}} = \massmat^{-1} \cdot (\ldots - \bm{D}\dot{\bm{q}} - \ldots) \]
Suppose we have damping only at the elbow: \(\bm{D} = \text{diag}(0, c_2)\). The damping term entering the inverse is: \[ -\bm{D}\dot{\bm{q}} = \begin{pmatrix} 0 \\ -c_2 \dot{q}_2 \end{pmatrix} \]
After multiplication by \(\massmat^{-1}\), the shoulder acceleration \(\ddot{q}_1\) receives a contribution from this elbow damping force, even though the shoulder has zero damping directly. Quantitatively: \[ \ddot{q}_1 = \frac{1}{\det(\massmat)} \left[ (\massmat^{-1})_{11} (-c_2 \dot{q}_2) + (\massmat^{-1})_{12} (-c_2 \dot{q}_2) \right] \]
The cross-term \((\massmat^{-1})_{12}\) couples the two joints. Damping at the elbow affects the shoulder via this off-diagonal path.
This is crucial: in a multi-joint chain, you cannot treat each joint’s dynamics in isolation. The mass matrix creates a network of inertial coupling. Damping at a distal joint propagates proximally.
Coupling Through Coriolis Effects
The Coriolis matrix \(\cormat\) also couples joints. Damping reduces velocity, which reduces Coriolis forces: \[ \cormat(\bm{q}, \dot{\bm{q}}) \dot{\bm{q}} \]
When \(\bm{D}\) reduces \(\dot{\bm{q}}\), the Coriolis term decreases, which indirectly affects accelerations everywhere. This is a second-order effect (damping \(\to\) velocity \(\to\) Coriolis \(\to\) acceleration) but it’s real.
Energy Cascade in Serial Chains
Consider a serial chain: shoulder \(\to\) elbow \(\to\) wrist \(\to\) club shaft. The kinetic energy flows from proximal (shoulder) to distal (club). At each joint, damping dissipates energy. Where does this dissipation hurt the most?
In a serial kinematic chain, damping at proximal (root-side) joints dissipates energy from the entire downstream chain. Damping at distal (tip-side) joints only affects that joint and its descendants. Quantitatively, the fractional energy loss is much larger for proximal damping than distal damping of equal coefficient magnitude.
Why? Because the proximal joint has higher angular velocity (relative to body-fixed rates) and moves a larger inertia. The power dissipated by damping is \(P_d = c \omega^2\). In a maximally extended arm, the shoulder rotates at roughly 7000\({}^\circ\)/s (approximately 122 rad/s), whereas the wrist rotates at roughly 3000\({}^\circ\)/s relative to the body (Nesbit 2005; Hume et al. 2005) (illustrative order-of-magnitude estimates; exact values depend on individual and swing phase). Even if both have the same damping coefficient \(c\), the shoulder dissipates approximately 17 times more power.
Constraint Coupling in Parallel Mechanisms
Now consider a two-handed grip on the club. The two hands are connected by the constraint that they must both hold the same club. This constraint can be written as: \[ \bm{J}_L(\bm{q}_L) \bm{v}_L = \bm{J}_R(\bm{q}_R) \bm{v}_R \] where subscripts \(L\), \(R\) denote left and right hands.
The constraint forces balance to keep the club in place: \[ \bm{J}_L^T \bm{f}_L + \bm{J}_R^T \bm{f}_R = 0 \]
Now, if the left arm has damping and the right does not, the left arm “resists” motion more. This resistance gets transmitted through the constraint forces, which then increase the damping effect on the right arm. The two-handed grip creates a damping coupling path that does not exist in a single-arm open chain.
Concretely, if left arm damping causes the left hand to lag behind, the constraint forces increase to pull it back, and this pulling force increases the effective damping on the right hand. The two arms are no longer independent; they are coupled through the constraint mechanism.
The Damping Paradox
Here is the apparent paradox: damping dissipates energy, which is bad for distance. But damping also stabilizes motion, which is good for control. How can we reconcile this?
Consider the linearized dynamics around an equilibrium: \(\dot{\bm{x}} = A\bm{x}\). The eigenvalues of \(A\) determine stability. If \(A\) has a term \(-\bm{D}\) (damping), then damping moves the eigenvalues into the left half-plane (negative real parts), making the equilibrium stable. Without damping, eigenvalues can lie on the imaginary axis (marginal stability) or right half-plane (instability).
Damping trades energy loss for stability.
Underdamped vs. Overdamped Joints
A joint can be classified by its damping ratio: \[ \zeta = \frac{c}{2\sqrt{km}} \] where \(c\) is the damping coefficient, \(k\) the stiffness, and \(m\) the effective mass.
\(\zeta < 1\) (underdamped): The joint oscillates as it returns to rest. Energy is dissipated over multiple cycles. Motion is fast but wiggly.
\(\zeta = 1\) (critically damped): The joint returns to rest in the shortest time without overshoot. Energy dissipation is optimal for this goal.
\(\zeta > 1\) (overdamped): The joint slowly creeps back to rest without oscillating. Motion is sluggish but stable.
Most human joints are underdamped. For example, tap your knee and notice how your leg oscillates slightly. The damping is not strong enough to eliminate the wobble completely. This is actually beneficial: underdamped systems are energetically efficient at some operating points. The oscillations can store and release energy in elastic elements (tendons, muscles).
During the golf swing, underdamping of the wrist allows it to cock and uncock in a near-resonant manner, storing elastic energy in the wrist extensors and releasing it at the right moment. If the wrist were overdamped (like a viscous oil-filled shock absorber), it would dissipate too much energy and would be “sluggish” — bad for clubhead speed.
Active Damping and Co-Contraction
Your nervous system can increase damping without increasing stiffness. This is done by co-contracting antagonist muscles. When the biceps and triceps both contract simultaneously, they stiffen the joint (increase stiffness) but also increase the internal friction (increase damping). This is called active damping or active impedance.
Why would the nervous system do this? For stability. During the transition from backswing to downswing, when the direction of motion reverses, damping helps prevent overshoot and wild oscillations. The brain increases co-contraction during this phase, sacrificing a bit of energy loss for a more controlled switch.
This connects to impedance control (Hogan 1985) from Chapter 27. Recall that you can write: \[ \bm{f} = \bm{K}(\bm{q} - \bm{q}_d) + \bm{D}(\dot{\bm{q}} - \dot{\bm{q}}_d) \] where \(\bm{K}\) and \(\bm{D}\) are task-space stiffness and damping. Both are under nervous system control. The brain does not choose \(\bm{K}\) and \(\bm{D}\) independently; they are correlated. Higher impedance environments (like hitting off hardpan) tend to cause increased co-contraction, which increases both \(\bm{K}\) and \(\bm{D}\).
The Loose Grip Debate
A long-standing debate in golf coaching: should the grip be loose or firm?
From a damping perspective, a firm grip increases the damping of the hand-club interface. The grip force compresses tissue, increasing friction. A loose grip reduces this damping, allowing the club to oscillate more freely.
A tighter grip () higher damping () more energy loss but better control. A looser grip () lower damping () more energy transmission to the club but less damping of impact shock.
The empirical data suggests that golfers find an intermediate grip pressure: tight enough to avoid the club slipping (which would be disastrous), but loose enough to allow some vibration that aids energy transfer. This is an implicit optimum balancing control and efficiency.
Friction Models for Joints
While damping typically refers to viscous resistance (proportional to velocity), friction typically refers to a dry contact force (direction-dependent, velocity-independent at leading order). The difference is subtle in practice: all joint friction has both viscous and dry components.
Coulomb Friction
The simplest friction model is Coulomb (kinetic) friction: the friction force is constant in magnitude, opposing the direction of motion: \[ f_{\text{friction}} = -\mu_k N \,\text{sign}(\dot{q}) \] where \(N\) is the normal force and \(\mu_k\) the coefficient of kinetic friction.
At low speeds or at reversal points (when \(\dot{q} = 0\)), static friction can be higher: \(\mu_s > \mu_k\).
Stiction and Reversal Points
At a reversal point — the instant when motion switches direction — static friction matters. The classic example in golf is the transition from backswing to downswing. At this instant, \(\dot{q}_{\text{shoulder}} = 0\). To initiate downswing, the shoulder must overcome static friction.
If static friction is very high, the joint becomes “sticky” — it resists initiating motion in the new direction. This is actually beneficial for control: it prevents the shoulder from drifting slightly backward when you intend to change direction. But if static friction is too high, you lose the smooth, continuous motion that characterizes an efficient swing.
Most human joints have \(\mu_s - \mu_k\) that is small enough that the stiction effect is minimal. The transition from backswing to downswing is not a sudden “sticking” event; it’s nearly continuous.
The Stribeck Effect
A more realistic friction model is the Stribeck effect, which captures the observation that friction decreases slightly with increasing velocity at very low speeds, before flattening out at higher speeds:
\[ f = -\left( \mu_s - (\mu_s - \mu_k) e^{-|\dot{q}|/v_s} \right) N \,\text{sign}(\dot{q}) \]
At \(\dot{q} = 0\): \(f = -\mu_s N\) (static friction). At \(\dot{q} \to \infty\): \(f \to -\mu_k N\) (kinetic friction). For \(0 < \dot{q} < v_s\) (very slow motion): \(f\) decreases smoothly from static to kinetic.
The Stribeck effect is subtle but important for joint dynamics at the reversal point. It explains why very slow movements (like address posture adjustments) feel “sticky,” while faster movements feel smoother.
LuGre Friction Model
For high-fidelity modeling (beyond the scope of this book), the LuGre model captures: 1. Stribeck effect. 2. Bristle deflection (microscopic stiction elements). 3. Hysteresis.
\[ \dot{z} = \dot{q} - \sigma_0 |{\dot{q}}| \frac{z}{g(v)} \] \[ f = \sigma_0 z + \sigma_1 \dot{z} + \sigma_2 \dot{q} \]
where \(z\) represents an internal “bristle deflection” state. This model is accurate for precision engineering but is overkill for biomechanics.
Friction Criticality at the Transition
The backswing-to-downswing transition is the moment when friction is most critical. At this moment: - The shoulder, elbow, and wrist are all at peak stretch (maximum elastic energy in muscles and tendons). - Velocities reverse, causing stiction effects to activate. - Co-contraction may increase damping for stability. - The direction reversal means the sign of friction forces must flip.
A skilled golfer executes this transition in \(\sim 20\) milliseconds, during which the kinetic energy of the backswing must be redirected into the downswing. Excessive friction at this moment would dissipate stored elastic energy. Insufficient friction (stiction) would allow slipping.
Empirically, most golfers find that their most consistent, repeatable swings occur when they have a smooth, unhurried transition — which is consistent with minimizing friction losses at the reversal point. Jerky transitions (common in amateurs) create brief moments of velocity reversal with high stiction, wasting energy.
Energy Dissipation Budget
Let’s add up the numbers. A typical amateur golfer generates approximately: - Total muscle energy per swing: 300–400 J - Kinetic energy at clubhead impact: 120–150 J - Missing energy: 150–280 J
Where does this energy go?
Sources of Dissipation
| Dissipation Source | Energy (J) | % of Total |
|---|---|---|
| Joint damping (all joints) | 40–60 | 13–15% |
| Tendon hysteresis | 20–30 | 6–8% |
| Muscle internal friction | 30–50 | 10–13% |
| Body segment soft tissue | 15–25 | 5–8% |
| Aerodynamic drag (arm/torso) | 10–15 | 3–5% |
| Ground deformation / contact friction | 5–15 | 2–5% |
| Grip/hand friction losses | 5–10 | 2–3% |
| Swing-to-impact transition inefficiency | 20–40 | 7–10% |
| Total dissipation | 145–245 | 48–65% |
Note: These are rough illustrative estimates based on simplified biomechanical models (Nesbit 2005; MacKenzie and Sprigings 2009); actual values depend on individual swing mechanics, body composition, and equipment and vary widely across studies.
Interpretation
The crucial insight: roughly half of the muscular energy generated does not reach the club. This is not a failure; it’s unavoidable. The question is whether the energy is dissipated efficiently (in a way that improves control and stability) or wastefully (in a way that reduces distance without improving anything).
Key observations:
Joint damping accounts for 13–15% loss. This is significant and not negligible. A \(\pm 50\%\) change in damping would shift distance by roughly 6–8%, all else being equal.
Tendon hysteresis is one of the largest dissipative sources. Yet tendons also store elastic energy. The net effect of tendon mechanics is beneficial: they act as elastic accumulators, releasing energy at optimal moments.
Body segment soft tissue wobble dissipates 5–8%. This is the “shake” of muscle, fat, and skin. Stronger individuals with less soft tissue wobble (i.e., low body fat) dissipate less here.
Aerodynamic drag is small in golf (unlike baseball). The arm and club move through air, but the drag is not the primary energy sink.
The swing-to-impact transition is a hidden source of loss. The kinetic energy of the arms and trunk must be redirected and partially transferred to the club. This is not 100% efficient. Good technique minimizes this loss; poor technique (e.g., casting the club early, losing lag) amplifies it.
Implications for Swing Modeling
Most biomechanical and golf models in the literature ignore joint damping. They solve the undamped equations of motion and use only muscle forces and gravity. How bad is this simplification?
When Damping Is Negligible
Damping effects are smallest when: 1. Velocities are low (address, takeaway, follow-through). 2. Motions are slow (a practice swing). 3. Joints are well-lubricated and moving in a smooth, extended manner.
In the address position, speeds are near zero, so damping forces are negligible. In the early takeaway, velocities are building up gradually, and inertial effects (mass matrix, Coriolis) dominate. Damping is a perturbation.
When Damping Is Critical
Damping effects are largest when: 1. Velocities are high (late downswing, especially the final \(\sim 100\) ms before impact). 2. Joints are transitioning (reversal points like mid-downswing swing changes). 3. Joints are moving against viscous resistance (e.g., wrist supination-pronation, which involves heavy tendon routing).
During the final 100 ms of the downswing, club speed is changing at a rate of approximately 50 m/s\(^2\) (Nesbit 2005) (illustrative: local acceleration; this is a typical-order estimate from simulation; depends on swing speed and phase; roughly 30–80 m/s\(^2\) is a plausible estimated range). At these speeds, a damping torque of even 1 Nm becomes significant.
Sensitivity Analysis
In this illustrative sensitivity analysis, we can estimate the effect of damping on predicted club head speed. A first-order approximation:
\[ \Delta v_{\text{club}} \approx -\frac{\partial v_{\text{club}}}{\partial D} \cdot \Delta D \]
Empirically, this derivative is approximately \(-0.02\) m/s per (Nm\(\cdot\)s/rad) of damping. Thus:
- \(50\%\) increase in damping (\(+5\) Nm\(\cdot\)s/rad): \(\Delta v \approx -0.1\) m/s, or \(\sim 2\) mph loss.
- \(50\%\) decrease in damping (\(-5\) Nm\(\cdot\)s/rad): \(\Delta v \approx +0.1\) m/s, or \(\sim 2\) mph gain.
For a driver swing (club speed roughly 80–90 mph, illustrative range for intermediate to advanced golfers (Broadie 2014; TrackMan 2023); depends on individual; typical values vary across player populations), a 2 mph change is about 5% of club speed. This is not negligible, but it’s not dominant either. Most of the sensitivity in club speed comes from muscular timing and force magnitude, not from damping.
However, in this simplified sensitivity picture, damping effects can amplify, especially if the downswing phase is extended (slow-motion downswing). In such cases, damping may account for 10–15% of the club speed variation.
Recommendations for Modelers
If modeling the full swing (address to follow-through): In this illustrative modeling recommendation, include a constant damping matrix with values from the table in Section 1.1.6. Use linear viscous damping; the added complexity of nonlinear friction is not justified by the accuracy gain for most purposes.
If modeling only the downswing or impact: Damping becomes more important. Use configuration-dependent damping \(\bm{D}(\bm{q})\) with higher coefficients when joints are flexed deeply.
If performing sensitivity analysis: Vary damping coefficients by \(\pm 50\%\) and report the effect on club speed and trajectory. This provides bounds on the unmodeled dynamics.
If comparing model predictions to experiment: Remember that damping is one of several energy sinks. Unmodeled losses (aerodynamic, ground friction, swing-to-impact inefficiency) might dwarf damping losses. Damping alone cannot reconcile model-experiment discrepancies if other losses are large.
For control simulations: Always include damping. Even though its energy cost is modest, damping dramatically improves stability of numerical simulations and realism of the control response.
Damping in the Shaft
Recall from Chapter 11 that the shaft is a distributed elastic beam. In addition to stiffness, shafts have material damping — the internal friction of the composite material itself.
Material Damping
When you flex a golf shaft and release it, the oscillations decay over time. This decay is due to material damping: hysteresis in the material’s stress-strain curve. The damping is frequency-dependent and material-dependent.
For an oscillating beam, the quality factor is: \[ Q = 2\pi \frac{\text{Peak energy stored}}{\text{Energy dissipated per cycle}} = \frac{\omega_0}{2\zeta \omega_0} = \frac{1}{2\zeta} \] where \(\zeta\) is the damping ratio. High Q (high stiffness, low damping) means oscillations persist. Low Q means oscillations decay quickly.
Typical Q-factors:
- Steel shaft: \(Q \approx 50--100\) (low damping, oscillations ring for many cycles).
- Graphite (carbon fiber) shaft: \(Q \approx 100--300\) (higher damping, oscillations decay slower).
- Composite (mixed fiber) shaft: \(Q \approx 50--150\) (intermediate).
The paradoxical result: carbon fiber, which is stiffer and lighter than steel, actually has lower material damping. So a graphite shaft rings longer after impact than a steel shaft of equal flex. Is this good or bad?
Lead-Lag Dynamics and Damping
The lead-lag angle \(\alpha\) (the angle between the shaft centerline and the club head velocity) is heavily influenced by shaft damping. A high-damping shaft (steel) is less likely to oscillate in lead-lag after impact. A low-damping shaft (graphite) may ring, causing transient vibrations that affect feel and feedback.
However, during the downswing (before impact), shaft damping is relatively unimportant. What matters is the elastic stiffness and mass distribution. Damping becomes critical after impact, during the deceleration phase.
Practical Implications
A stiffer, higher-damping shaft (like a stiff steel shaft) provides: - Less vibration after impact (more “solid” feel). - Less oscillation during swing. - Slightly more energy dissipated (lower ball speed, all else equal).
A lighter, lower-damping shaft (like a lightweight graphite) provides: - More vibration after impact (more “feedback”). - Slightly more efficient energy transfer during swing. - Higher ball speed, especially for slower swing speeds (where the golfer benefits from the shaft’s energy return during the downswing).
Modern drivers use relatively low-damping graphite shafts precisely because the efficiency gain (higher ball speed for a given muscular effort) outweighs the aesthetic loss of transient vibrations.
Reality: Steel shafts are more heavily damped, so they vibrate less after impact. But during the downswing, the difference in oscillation is minimal if the shaft is loaded predominantly in bending (not torsion). The “control” benefit of steel is more about tradition, feel, and perhaps placebo than about physics. Graphite shafts can be equally controllable if the golfer has proper timing and technique.
The Control-Affine Perspective on Damping
In the control-affine framework, damping modifies the drift field. Recall: \[ \dot{\mathbf{x}} = f_0(\mathbf{x}) + G(\mathbf{x})\control \]
When we add damping, the drift field becomes: \[ f_{\text{damped}}(\mathbf{x}) = f_0(\mathbf{x}) - \bm{D}_{\text{eff}} \dot{\bm{q}} \]
The effective damping \(\bm{D}_{\text{eff}}\) is not simply the sum of viscous and Coulomb components. It includes: 1. Viscous damping from joints and muscles. 2. Implicit damping from the Coriolis terms (reducing velocity reduces Coriolis forces). 3. Configuration-dependent effects (soft tissue compression, muscle routing changes).
The drift field represents the “natural” motion of the system under gravity and inertia, without any muscular control. When damping is added, the drift field is damped: the velocity naturally decays, and extreme accelerations are limited. This is why damping stabilizes the system.
Equivalently, the control field \(G(\mathbf{x})\) must work against the damping to accelerate the joints. The control inputs must provide enough torque to overcome damping. This is one reason why the transition from backswing to downswing requires a sudden increase in muscular effort: the brain must overcome damping forces to reverse the direction of motion.
Summary: The Damping Principle
Damping is not a bug in the golf swing; it’s a feature. The right amount of damping at the right joints: - Stabilizes the motion, preventing wildly unstable oscillations. - Provides feedback and proprioceptive cues. - May limit peak accelerations, which biomechanical reasoning suggests could help reduce joint loading. - Allows the nervous system to implement impedance control, modulating stiffness and damping together.
The challenge is tuning damping for the task. Too much, and you lose efficiency. Too little, and you lose stability. A good golf swing implicitly balances these two objectives through unconscious learning and adaptation.
Chapter Exercises
Damping Coefficient Estimation. Suppose a golfer’s wrist has a damping coefficient of \(c_w = 0.1\) Nm\(\cdot\)s/rad. During the downswing, the wrist angular velocity is \(\dot{q}_w = 50\) rad/s. What is the damping torque? If the maximum wrist torque the muscles can provide is 20 Nm, what percentage of the muscular torque is consumed by damping?
Energy Dissipation in a Two-Joint Chain. Consider a simplified shoulder-elbow system. The shoulder has angular velocity \(\dot{q}_1 = 10\) rad/s with damping coefficient \(c_1 = 0.5\) Nm\(\cdot\)s/rad. The elbow has \(\dot{q}_2 = 30\) rad/s with \(c_2 = 0.15\) Nm\(\cdot\)s/rad. Calculate the total power dissipated by damping: \(P_d = \sum c_i \dot{q}_i^2\). If the total muscular power output is 400 W, what fraction is lost to damping?
Damping Ratio and Step Response. A wrist joint has mass \(m = 0.05\) kg\(\cdot\)m\(^2\), stiffness \(k = 5\) Nm/rad, and damping \(c = 0.2\) Nm\(\cdot\)s/rad. Compute the natural frequency \(\omega_n = \sqrt{k/m}\), the critical damping \(c_c = 2\sqrt{km}\), and the damping ratio \(\zeta = c/c_c\). Is the joint underdamped, critically damped, or overdamped? Estimate the settling time (time to reach 2% of steady-state value) using \(t_s \approx 4 / (\zeta \omega_n)\).
Coulomb vs. Viscous Friction. A shoulder joint experiences Coulomb friction with coefficient \(\mu_k = 0.1\) and normal force \(N = 50\) N, giving a constant friction torque of 5 Nm. For comparison, viscous damping at the shoulder is \(c_s = 0.5\) Nm\(\cdot\)s/rad. At what shoulder angular velocity is the viscous damping torque equal to the Coulomb friction torque? For velocities below this threshold, which damping model dominates? Above it?
Configuration-Dependent Damping. Suppose the wrist damping increases with wrist flexion angle according to \(c_w(q_w) = 0.1 + 0.02 q_w\) Nm\(\cdot\)s/rad, where \(q_w\) is in radians. If the wrist flexes from \(q_w = 0\) rad to \(q_w = 1\) rad during the downswing, and the wrist velocity is constant at \(\dot{q}_w = 50\) rad/s, compute the instantaneous damping torque at \(q_w = 0\), \(q_w = 0.5\), and \(q_w = 1\) rad. How much does the damping torque change over this range?