Applications of Derivatives

Introduction

Differentiation is one of the principal mathematical tools used to describe change. In mechatronics engineering, quantities such as position, temperature, voltage, force, energy consumption, and manufacturing cost rarely remain constant. A derivative measures how rapidly one quantity changes with respect to another and therefore supports prediction, control, design, and optimization.

If a variable y depends on x, its derivative is written dy/dx. Geometrically, it is the slope of the tangent to the graph of y against x. Physically, it is a local rate of change. The derivative is meaningful only when the variables and units are specified. For example, if distance is measured in metres and time in seconds, ds/dt has units of metres per second.

This chapter develops applications of derivatives to motion, related rates, curve analysis, measurement uncertainty, sensitivity, numerical root-finding, and constrained engineering optimization.

Learning Objectives

By the end of this chapter, you should be able to:

• interpret first and second derivatives in engineering contexts;
• calculate velocity and acceleration from position functions;
• solve related-rates problems using implicit differentiation;
• analyze curves using critical points, concavity, and asymptotes;
• estimate error propagation and sensitivity;
• apply Newton’s method and assess its limitations;
• formulate and solve constrained design and resource-allocation problems;
• evaluate engineering results using units, assumptions, and physical feasibility.

1. Derivatives as Engineering Rates

For a differentiable function y = f(x), the derivative at x is defined by

Here, h is an increment in the independent variable. The derivative exists when this limit exists. The derivative may be interpreted as:

1. A geometric slope: the slope of the tangent line.
2. A physical rate: the change in y per unit change in x.
3. A local linear model: for a small change ∆x,

The approximation assumes that f is differentiable and that ∆x is sufficiently small. The corresponding tangent-line approximation is

The second derivative measures how the first derivative changes:

In a mechanical system, a positive second derivative may indicate increasing velocity, while in a cost model it may indicate increasing marginal cost.

Common derivative rules

For constants a, n, and differentiable functions u and v:

The last expression is the chain rule. It is especially important when a sensor output depends on an intermediate physical quantity.

2. Position, Velocity, and Acceleration

Let s(t) denote position as a function of time t. The instantaneous velocity is

and the instantaneous acceleration is

If s is in metres and t is in seconds, v is in metres per second and a is in metres per second squared. Velocity is signed: its sign indicates direction relative to the chosen coordinate axis. Speed is the non-negative magnitude |v(t)|.

Average velocity over an interval [t₁,t₂] is

Average velocity and instantaneous velocity should not be confused. A vehicle can have zero average velocity over a trip while having nonzero instantaneous velocity throughout.

Worked Example 1: Motion of a robotic actuator

The position of a linear actuator is modeled by s(t)=0.4t³−1.2t²+2t metres, where t is in seconds.

Differentiate:

At t = 2 s,

Carefully, 1.2(4)=4.8, 2.4(2)=4.8, so v(2)=2 m/s. Also,

Thus the actuator is moving in the positive direction at 2 m/s and its velocity is increasing at 2.4 m/s². The displayed intermediate expression is more safely read as v(2)=4.8−4.8+2=2; this illustrates why arithmetic checks are important.

The actuator is momentarily at rest when v(t) = 0. The quadratic has discriminant

(−2.4)²−4(1.2)(2)=−3.84, which is negative. Therefore it never stops under this model.

3. Related Rates

A related-rates problem involves several changing quantities linked by an equation. The standard method is:

1. Define variables and units.
2. Write the geometric or physical relation.
3. Differentiate with respect to time.
4. Substitute the known instantaneous values.
5. Solve for the required rate and check its sign and units.

Substitution should normally occur after differentiation. Substituting too early can incorrectly treat a changing quantity as constant.

Worked Example 2: Robotic arm geometry

The end effector of a planar arm is at coordinates (x,y), with distance from the base

where L is constant. If dx/dt=0.30 m/s, x = 0.40 m, and y = 0.30 m, find dy/dt.

Differentiate with respect to t:

Hence,

Substitution gives

The negative sign means that y is decreasing at 0.40 m/s. This result assumes a rigid arm and motion constrained to a circle of constant radius.

Related rates also occur in electrical and thermal systems. If capacitor charge is q=CV and capacitance C is constant, then dq/dt=C dV/dt. If C changes, the product rule is required: dq/dt=C dV/dt+V dC/dt.

4. Curve Analysis and Engineering Interpretation

Curve analysis helps engineers identify operating limits, turning points, trends, and regions of diminishing or increasing performance.

Critical points

For f defined on an interval, a critical number is a point c where f′(c)=0 or where f′ is undefined, provided c lies in the domain. A critical point is not automatically a maximum or minimum. It must be classified using additional evidence.

The first-derivative test uses the sign of f′:

• positive to negative: local maximum;
• negative to positive: local minimum;
• no sign change: neither, or a stationary inflection.

The second-derivative test states that if f′(c)=0 and f′′(c)>0, f has a local minimum at c; if f′′(c)<0, it has a local maximum. If f′′(c)=0, the test is inconclusive.

Concavity and inflection

A graph is concave upward where f′′(x)>0 and concave downward where f′′(x)<0. An inflection point is a point where concavity changes. It is not sufficient merely for the second derivative to equal zero.

Absolute extrema

On a closed interval [a,b], a continuous function attains an absolute maximum and minimum. To find them, evaluate f at all interior critical points and at the endpoints. Engineering design variables frequently have such bounds, so endpoint checking is essential.

Asymptotes and model validity

A vertical asymptote can arise when a denominator approaches zero. Horizontal or oblique asymptotes describe long-range behavior. An asymptote may signal a physical singularity, but it may also indicate that a mathematical model is being used outside its valid range. For example, a linearized sensor equation may predict unrealistic values beyond its calibration interval.

Worked Example 3: Efficiency model

Suppose the normalized efficiency of a mechanism is

where x is a dimensionless load index and η is dimensionless. Then

Setting η′(x) = 0 gives x = 3. Also, η′′(x) = −0.04 < 0, so x = 3 is a local maximum. Check endpoints:

η(0)=0.80, η(3)=0.80+0.36−0.18=0.98, and η(4)=0.96.

Thus the maximum on the specified interval is 0.98 at x = 3. The model predicts decreasing efficiency after that point, but the conclusion is valid only within the stated interval and model assumptions.

5. Error Propagation and Sensitivity

Measurements are never exact. Derivatives provide first-order estimates of how input uncertainty affects an output.

If y=f(x) and x has a small absolute uncertainty ∆x, then

The relative sensitivity of y to x is often represented by

For small changes,

This dimensionless quantity indicates the percentage change in y caused by a one-percent change in x. A large magnitude indicates high sensitivity.

For multiple independent variables, a conservative first-order worst-case estimate is

If independent random standard uncertainties u(xᵢ) are used instead, the root-sum-square estimate is

The root-sum-square expression assumes small uncertainties, a differentiable model, and mutually independent random inputs. Correlated inputs require covariance terms.

Worked Example 4: Area measurement

A circular sensor target has diameter d = 50.0 mm with uncertainty u(d) = 0.2 mm. Its area is

Therefore,

At d = 50.0 mm,

The nominal area is A=π(50.0)²/4≈1963.5 mm². The relative uncertainty is approximately 15.7/1963.5≈0.008, or 0.8%. Since A is proportional to d², the relative uncertainty is also approximately twice the relative diameter uncertainty.

6. Newton’s Method

Many engineering equations cannot be solved conveniently in closed form. Newton’s method approximates a root of f(x)=0 using the tangent line at a current estimate xₙ:

The method requires f to be differentiable near the root and requires f′(xₙ)≠0. A reasonable initial estimate is important.

Worked Example 5: Calibration equation

Find a root of f(x)=x²−2 using x₀ = 1.5. Since f′(x)=2x,

Next,

A further iteration gives x₃ ≈ 1.581139, which is close to √2≈1.414214?

Correction: the arithmetic exposes an inconsistency: for f(x)=x²−2, the square root is approximately 1.414214, not 1.581139. Recalculate the first step: f(1.5)=2.25−2=0.25, so x₁=1.5−0.25/3=1.416667. Then x₂≈1.414216 and x₃ ≈ 1.414214. This example demonstrates the necessity of checking both the formula and arithmetic.

A practical stopping rule is based on either successive estimates, |xₙ₊₁−xₙ|<ε, or residual, |f(xₙ)|<ε_f, with tolerances selected according to engineering accuracy. Newton’s method can fail through divergence, cycling, convergence to an unintended root, or division by a very small derivative. Safeguards include bracketing, damping, and a maximum iteration count.

7. Optimization and Constrained Design

Optimization seeks the best feasible value of an objective function. In engineering, the objective may be cost, mass, energy, time, error, or negative performance. A general constrained problem is

A valid solution must satisfy all constraints. A local optimum is best in a neighborhood; a global optimum is best over the entire feasible region.

One-variable procedure

For a differentiable objective F(x) on a feasible interval:

1. find interior points where F′(x)=0;
2. include points where F′ is undefined;
3. include all feasible endpoints and active constraint boundaries;
4. compare objective values;
5. verify physical feasibility and units.

Lagrange multipliers

For maximizing or minimizing F(x,y) subject to g(x,y)=c, define

or equivalently,

This condition applies at a smooth constrained interior optimum where the gradient of the constraint is nonzero. It identifies candidates, not automatically the final optimum; candidates must be compared and checked against boundaries.

Case Study: Design of a rectangular inspection station

Background. An automated inspection station requires a rectangular work area with fixed perimeter 20 m. The width is x and the length is y. The design goal is to maximize usable area, subject to 2 m ≤ x ≤ 8 m and x,y > 0.

Problem. Determine the dimensions giving maximum area.

Analysis. The perimeter constraint is

so y = 10 − x. The area is

The derivative is A′(x)=10−2x. Setting it to zero gives x=5 m. Since A′′(x)=−2<0, this is a maximum. It is feasible because 2 ≤ 5 ≤ 8. The corresponding length is y=5 m and area is 25 m². Endpoint areas are A(2)=16 m² and A(8)=16 m², confirming the interior maximum.

Approach and engineering interpretation. The mathematical result produces a square because, with fixed perimeter, equal dimensions maximize area. In a real station, however, camera field of view, conveyor orientation, access clearance, structural supports, and safety regulations may impose additional constraints. The unconstrained mathematical optimum is therefore a starting design, not an automatic final specification.

Outcome. The idealized design is 5 m × 5 m. If a required conveyor makes x at least 6 m, the constrained optimum must be reconsidered; the best feasible point may occur at the active boundary x=6 m.

Learning points. Optimization requires an objective, explicit constraints, candidate comparison, and engineering validation. A mathematically optimal design can be operationally unsuitable if important constraints were omitted.

Resource trade-offs

Suppose a controller’s total cost is C(p,q), where p is processor capacity and q is sensor resolution. Increasing either variable may improve performance but increase cost. A constrained budget or performance requirement creates a feasible set. Derivatives quantify marginal trade-offs: ∂C/∂p is the marginal cost of processor capacity, while ∂P/∂q measures the marginal performance benefit of resolution. Decisions should compare marginal benefit with marginal cost rather than relying only on absolute values.

8. Professional Practice, Limitations, and Contemporary Tools

Derivative-based results depend on model quality. Common limitations include:

• noisy measurements, which can make numerical differentiation unstable;
• discontinuities, saturation, backlash, and friction that violate smooth models;
• omitted constraints such as safety, manufacturability, or thermal limits;
• local rather than global optimization;
• inappropriate extrapolation beyond calibration data;
• rounding errors and poorly scaled variables.

In digital control and data acquisition, derivatives are often approximated by finite differences. For sample interval ∆t,

Small ∆t reduces truncation error but can amplify sensor noise. Filtering, appropriate sampling, dimensional analysis, and validation against physical limits are therefore essential.

Computer algebra, numerical solvers, and AI-assisted engineering software can differentiate models, search design spaces, and identify candidate solutions. They do not replace engineering judgment. Users must verify assumptions, units, boundary conditions, numerical convergence, and the provenance of data. Safety-critical optimization must remain auditable, with human review and documented acceptance criteria.

Chapter Summary

Derivatives describe local change and provide a common language for motion, sensing, design, and performance analysis. The first derivative gives rates such as velocity, marginal cost, and sensitivity; the second derivative describes acceleration or curvature. Related-rates problems are solved by differentiating governing relations before substituting instantaneous values. Curve analysis uses critical points, derivative signs, concavity, endpoints, and asymptotes, while recognizing model limitations. First-order error propagation estimates uncertainty and sensitivity, with root-sum-square methods appropriate for independent random uncertainties. Newton’s method uses tangent-line corrections to solve nonlinear equations but requires suitable starting values and safeguards. Engineering optimization combines objective functions with explicit physical and resource constraints; candidate solutions must be compared and validated against real operating requirements.

Key Terms

Derivative: Instantaneous rate of change of one variable with respect to another.

Tangent: Line that locally approximates a differentiable curve.

Velocity: Time derivative of position.

Acceleration: Time derivative of velocity.

Critical point: Domain point where a derivative is zero or undefined.

Local maximum: Value greater than nearby feasible values.

Local minimum: Value less than nearby feasible values.

Concavity: Direction in which a curve bends, determined by the second derivative.

Inflection point: Point where concavity changes.

Related rates: Rates linked through a common equation and differentiated with respect to time.

Sensitivity: Change in an output caused by change in an input.

Error propagation: Estimation of output uncertainty from input uncertainty.

Worst-case uncertainty: Sum of absolute first-order contributions.

Root-sum-square uncertainty: Uncertainty estimate for independent random inputs.

Newton’s method: Iterative tangent-based method for solving f(x)=0.

Objective function: Quantity to be minimized or maximized.

Constraint: Equality, inequality, or bound restricting feasible designs.

Lagrange multiplier: Auxiliary variable used to locate smooth constrained extrema.

Feasible set: All points satisfying the constraints.

Marginal value: Derivative measuring the incremental effect of a variable.

Review Questions

Short Answer Questions

1. Distinguish instantaneous velocity from average velocity.
2. Why should substitution usually be performed after differentiating in a related-rates problem?
3. What information does the second derivative provide about a function?
4. State the first-order absolute error-propagation formula for y=f(x).
5. Why can Newton’s method fail even when f is differentiable?

Long Answer Questions

1. Explain how position, velocity, and acceleration are related, including units and sign conventions.
2. Describe a complete procedure for finding absolute extrema on a closed interval.
3. Explain the assumptions behind root-sum-square uncertainty propagation.
4. Derive the Lagrange-multiplier conditions for a two-variable equality-constrained problem.
5. Discuss why mathematical optimality does not necessarily imply engineering suitability.

Analytical / Application Questions

1. A motor shaft position is θ(t)=2t³−5t²+4t radians. Find angular velocity and angular acceleration, and determine their values at t=1 s.
2. A circular tank has radius r and volume V=πr²h, where h is constant. Derive dV/dt in terms of r and dr/dt, and interpret the units.
3. A manufacturer has a fixed material perimeter for a rectangular housing but requires one side to be at least 6 cm. Formulate the area-maximization problem, identify the unconstrained optimum, and explain how the lower-bound constraint could alter the design.

Multiple Choice Questions with Answers

Question 1

If s(t) is position in metres and t is time in seconds, what are the units of s′′(t)?

A. m

B. m/s

C. m/s²

D. s/m

Correct Answer: C. m/s²

Question 2

For f(x)=x³−4x, which expression is f′(x)?

A. x²−4

B. 3x²−4

C. 3x−4

D. x³−4

Correct Answer: B. 3x²−4

Question 3

If a function changes from increasing to decreasing at a critical point, that point is a

A. local minimum

B. local maximum

C. vertical asymptote

D. stationary inflection necessarily

Correct Answer: B. local maximum

Question 4

For the relation x²+y²=L² with constant L, which equation is obtained by differentiating with respect to time?

A. x dx/dt + y dy/dt = L

B. 2x dx/dt + 2y dy/dt = 0

C. x²+y²=0

D. dx/dt+dy/dt=L²

Correct Answer: B. 2x dx/dt + 2y dy/dt = 0

Question 5

If y=3x² and x changes by approximately 0.01 at x=2, the first-order change in y is approximately

A. 0.01

B. 0.06

C. 0.12

D. 1.20

Correct Answer: C. 0.12

Question 6

Which condition is required for the standard Newton iteration at xₙ?

A. f′(xₙ)=0

B. f′(xₙ)≠0

C. f(xₙ)=1 necessarily

D. xₙ must be an integer

Correct Answer: B. f′(xₙ)≠0

Question 7

For A(x)=10x−x², the interior critical point is

A. x=0

B. x=2

C. x=5

D. x=10

Correct Answer: C. x=5

Question 8

A positive second derivative at a stationary point indicates a local minimum when

A. f′′(c)>0

B. f′′(c)<0

C. f′(c)>0

D. f(c)=0

Correct Answer: A. f′′(c)>0

Question 9

The sensitivity S=(x/y)(dy/dx) is

A. always measured in metres

B. dimensionless when x and y have consistent units

C. always negative

D. equal to the second derivative

Correct Answer: B. dimensionless when x and y have consistent units

Question 10

In a constrained optimization problem, a feasible point is one that

A. has the largest objective value only

B. satisfies all stated constraints

C. always has zero gradient

D. lies outside the design bounds

Correct Answer: B. satisfies all stated constraints