What is local stiffness in simple terms?

Local stiffness is the slope of the force-displacement curve at one specific operating point. It is the spring constant you would measure if you made a very small push or pull from that point.

Can I use this method for a spring that is not described by a polynomial?

Yes. You only need the derivative at x₀. If you have a measured curve, estimate the slope by drawing the tangent line or by fitting a small polynomial around x₀ and differentiating.

How small does Δx need to be for the approximation to be good?

There is no universal number. A rule of thumb: keep Δx less than 10% of x₀, and check that the second derivative term (0.5 × k'(x₀) × Δx²) is less than 5% of the linear term. For the example above, that condition was satisfied.

What units does the stiffness come out in?

Stiffness k is always in newtons per meter (N/m), Δx in meters (m), and force change ΔF in newtons (N). These are the SI units defined by the BIPM SI Brochure 9th edition.

Why not just use the secant stiffness between two points?

The secant stiffness gives the average over an interval, not the instantaneous response at the operating point. For small Δx, the secant and tangent are nearly the same; for larger Δx, the secant is the better choice because it accounts for the nonlinearity over the whole interval.

Estimate Spring Force From a Local Stiffness Curve

The Wrong Assumption: A Single Stiffness Number Works For Any Displacement

Most engineers treat a spring's stiffness as a single constant, 200 N/m, say, and multiply by total displacement to get force. That works only for linear springs. For a nonlinear spring, the force-displacement curve changes slope as you push or pull farther. To estimate spring force from a local stiffness curve, you use the tangent slope at the exact operating point you are working near. That local slope, called the tangent stiffness, gives a linear approximation for a small change in displacement. It is the only safe way to predict a force increment when the spring behaves nonlinearly.

How The Tangent Estimates A Small Force Change Near One Operating Point

The tangent to a nonlinear force-displacement curve at a point x₀ is the line that just touches the curve at that point. Its slope is the derivative dF/dx evaluated at x₀, defined as the local stiffness k(x₀). For a small displacement Δx away from x₀, the force change ΔF is approximately k(x₀) × Δx. This is a first-order Taylor series approximation of the nonlinear function F(x) around x₀. The approximation treats the curve as locally straight, which is valid only when Δx is small relative to the curve's curvature.

Use the linear approximation calculator on this site to compute k(x₀) from your spring's force-displacement equation, then multiply by Δx. The result is an estimate, not an exact value. The nonlinear terms you dropped are proportional to (Δx)² and higher powers; those terms grow as Δx increases.

Estimate Spring Force From A Local Stiffness Curve: Worked Example

Consider a nonlinear spring with force given by F(x) = 200x + 500x³, where x is in meters and F is in newtons. The operating point is x₀ = 0.02 m. The tangent stiffness is the derivative at that point: dF/dx = 200 + 1500x². At x₀ = 0.02 m, that is 200 + 1500 × (0.0004) = 200 + 0.6 = 200.6 N/m.

For a small displacement Δx = 0.001 m (one millimeter), the linear estimate gives ΔF ≈ 200.6 × 0.001 = 0.2006 N. The actual force change, calculated from the full nonlinear equation, is [200×0.021 + 500×(0.021)³] − [200×0.02 + 500×(0.02)³] = [4.2 + 500×0.000009261] − [4 + 500×0.000008] = [4.2 + 0.0046305] − [4 + 0.004] = 4.2046305 − 4.004 = 0.2006305 N. The error is 0.0000305 N, or about 0.015%, negligible for almost any engineering purpose.

Calculation Steps

To carry out the calculation yourself, follow these steps:

Step 1: Obtain the force-displacement function. You need the explicit equation F(x) for your spring. If you only have a measured curve, fit a polynomial or spline to the data points, then differentiate analytically.

Step 2: Identify your operating point x₀. This is the displacement where the spring currently sits, the preload point in a mechanism, the equilibrium position in a vibration system, or the static deflection under a constant load.

Step 3: Compute k(x₀) = dF/dx at x₀. Use the derivative formula. For a polynomial F(x) = a₀ + a₁x + a₂x² + a₃x³ + ..., the derivative is a₁ + 2a₂x + 3a₃x² + ... Evaluate at x₀.

Step 4: Choose Δx, the small displacement you want the force change for. Keep Δx small, typically less than 10% of x₀ or less than the radius of curvature of the force-displacement curve, whichever is smaller.

Step 5: Compute ΔF ≈ k(x₀) × Δx. The result is in newtons when k is in N/m and Δx in meters.

Tangent Stiffness Values for Common F(x) Forms

F(x) Form	k(x) = dF/dx	Example At x₀ = 0.02 m
Linear: k₁x	k₁	k₁ = 200 N/m
Hardening: k₁x + k₂x³	k₁ + 3k₂x²	k₁=200, k₂=500: 200.6 N/m
Softening: k₁x − k₂x³	k₁ − 3k₂x²	k₁=200, k₂=500: 199.4 N/m
Cubic: k₃x³	3k₃x²	k₃=1000: 1.2 N/m

Why the Method Works

The tangent approximation fails when Δx is no longer small relative to the curvature of the force-displacement curve. The failure case from the research is instructive. For a hardening spring with F(x) = 100x + 1,000,000x³, at x₀ = 0.01 m, the tangent stiffness is k(0.01) = 100 + 3×1,000,000×(0.01)² = 100 + 300 = 400 N/m. For Δx = 0.005 m, the linear estimate gives ΔF = 400 × 0.005 = 2 N. The actual ΔF is [100×0.015 + 1,000,000×(0.015)³] − [100×0.01 + 1,000,000×(0.01)³] = [1.5 + 3.375] − [1 + 1] = 4.875 − 2 = 2.875 N. The error is 0.875 N, an error of 30%. At Δx = 0.01 m (equal to x₀ itself), the error would be even larger.

This failure is systematic: the cubic term introduces a (Δx)³ contribution that the linear approximation completely ignores. The relative error grows as the ratio of k₂x₀² to k₁ increases. For the worked example above (200x + 500x³ at x₀ = 0.02 m), the nonlinear term contributed only 0.6 N/m out of 200.6 N/m, so the linear approximation was excellent. For the failure case, the cubic term contributed 300 N/m out of 400 N/m, the spring was dominated by nonlinearity even at the operating point.

What To Do When The Approximation Fails

When Δx is large or the spring is strongly nonlinear, you cannot use the tangent line. Compute the exact ΔF from the full F(x) equation instead. If you do not have an equation, run a physical test at the actual displacement you need. A common real-world example is a rubber shock absorber: its force-displacement curve is highly nonlinear, and the tangent stiffness at the static sag point may be meaningless for a large bump that compresses the shock by several inches.

Another approach is to use a piecewise linear model: break the force-displacement curve into segments, compute the average stiffness over each segment, and use that for displacements within the segment. This is how many finite element codes handle material nonlinearity.