What is the difference between a lumped model and a continuous model?

A lumped model treats a pile as a series of discrete masses connected by springs and dashpots, while a continuous model treats the pile as a continuous medium through which stress waves propagate in real time. The distinction matters because pile driving involves dynamic wave phenomena that a lumped approach can only approximate, whereas a continuous formulation captures the physics directly. The sections below unpack each model, compare their accuracy, and explain when each approach fits your project.

How does a lumped model represent a pile system?

A lumped model divides the pile into a finite number of discrete segments, each represented as a concentrated mass. Adjacent masses are connected by springs that simulate axial stiffness and dashpots that simulate damping. Soil resistance is applied at each mass node as an additional spring-dashpot element. The system is then solved using numerical time integration.

The classic example is the Smith model, which combines an elastic-plastic slider with a viscous dashpot at each soil interaction point. This formulation is straightforward to implement and has been widely used in pile driving analysis for decades. Because the pile is broken into a manageable number of elements, the computational setup is relatively simple and the input parameters are easy to understand intuitively.

However, lumped models carry an important limitation: the accuracy of the solution depends on the chosen time step for numerical integration. If the time step is too large relative to the element size and wave speed, numerical errors accumulate. This means that the discretization scheme and integration step must be selected carefully, and the results should always be checked for sensitivity to these numerical choices.

How does a continuous model represent a pile system?

A continuous model treats the pile as a physical medium with distributed mass, stiffness, and cross-sectional properties along its full length. Rather than lumping properties into nodes, it solves the wave equation continuously, tracking how stress waves propagate, reflect, and transmit through the pile as a function of both time and position.

The Method of Characteristics (MoC) is the approach used in continuous wave equation analysis. It traces the physical paths along which wave information travels through the pile. At boundaries where pile geometry changes, material properties shift, or soil resistance acts, the incoming wave is partially reflected and partially transmitted according to equilibrium and continuity conditions. This means that impedance changes, varying cross-sections, and localized soil reactions are all captured as part of the wave propagation physics rather than as approximations.

The MoC has a long history in foundation engineering. In the Netherlands, it was developed specifically for foundation piles, with early contributions adding shaft friction and toe resistance to the original free-rod formulation. This development line forms the foundation of the AllWave software package, which uses the MoC for both impact driving (AllWave-PDP) and vibratory driving (AllWave-VDP). A practical advantage of this formulation is that it does not suffer from the time-step sensitivity that affects lumped-mass finite-difference solutions, making it numerically more stable for this class of problem.

What are the practical differences in accuracy between the two models?

The continuous model is generally more accurate for problems where wave propagation details matter, because it directly solves the physics of stress wave travel rather than approximating it through discrete nodes. The lumped model can produce acceptable results when the pile is relatively uniform and the time step is chosen carefully, but it becomes less reliable as pile complexity increases.

Several practical factors drive the accuracy gap:

Impedance changes: When pile cross-section or material changes along the length, a continuous model captures the resulting reflections precisely. A lumped model can only approximate these effects through the chosen discretization.
Numerical stability: Lumped finite-difference solutions are sensitive to the integration time step. The MoC does not carry this same numerical instability risk for one-dimensional wave propagation.
Spatial resolution: A continuous model resolves stress distributions as a function of depth at any point in time. A lumped model provides values only at the discrete node locations.
Soil interaction detail: Both model types can incorporate elastic-plastic soil resistance with damping, but the continuous formulation allows soil reactions to be distributed more realistically along the pile shaft.

For drivability predictions and signal matching in dynamic load testing, the continuous approach provides a more direct representation of the underlying physics, which translates into more reliable results when pile geometry is complex or soil conditions vary significantly with depth.

When should a lumped model be used instead of a continuous model?

A lumped model is appropriate when the pile is geometrically simple, soil conditions are relatively uniform, and a quick estimate is acceptable. For preliminary assessments, educational purposes, or situations where computational resources are limited, a well-constructed lumped model can provide useful first-order results without the overhead of a full continuous wave equation analysis.

Lumped models are also reasonable when the frequency content of the loading is low relative to the pile’s natural frequencies, meaning that the discrete mass representation does not introduce significant error. In practice, this applies more readily to short, stiff piles than to long, slender offshore piles where wave travel time is significant.

You should move to a continuous model when any of the following apply:

The pile has variable cross-sections, wall thickness changes, or material transitions along its length
You need accurate stress distributions for structural checks, including tension and compression limits during driving
The project involves vibratory driving prediction or impact driving of long offshore piles where wave reflections are complex
Signal matching against measured force and velocity records is required
Sensitivity analysis demands a numerically stable platform that does not introduce time-step artifacts

How do lumped and continuous models handle soil resistance differently?

Both model types represent soil resistance through combinations of static capacity, stiffness (quake), and damping, but they differ in how these components are distributed along the pile and how they interact with the wave solution. In a lumped model, soil resistance is concentrated at discrete nodes. In a continuous model, resistance can be distributed continuously and interacts directly with the propagating wave at every point along the shaft.

Soil resistance in pile driving has two components: shaft resistance and toe resistance. Each must be modelled not only for maximum capacity but also for how resistance develops as a function of displacement and velocity. The classic Smith formulation uses an elastic-plastic slider combined with a viscous dashpot, which is a reasonable starting point but has known limitations when soil behaviour is more complex.

Research into dynamic shaft friction in sand shows that resistance can depend separately on acceleration, velocity, and displacement, which means a single viscous damping parameter does not always capture the full dynamic soil response. Remoulding during repeated hammer blows is another factor: soil resistance during installation can change over a series of impacts even when other model parameters remain constant. This is directly relevant to soil modelling for pile driving, because the resistance during installation is not the same quantity as the long-term static foundation capacity.

A continuous model handles these complexities more naturally because the wave solution at each depth step interacts with the local soil model at that point, rather than aggregating resistance into a node that represents a finite length of pile. For projects where accurate separation of shaft and toe contributions matters, such as in signal matching for dynamic load testing, this distributed treatment produces more physically meaningful results.

How We Help with Pile Driving Modelling

We apply continuous wave equation analysis as the foundation of our drivability and pile testing work, using the Method of Characteristics to capture the full complexity of stress wave behaviour in real pile systems. Our approach covers both impact and vibratory installation scenarios.

Specifically, we support your project with:

Drivability predictions using AllWave-PDP for impact driving and AllWave-VDP for vibratory driving, both built on the continuous MoC formulation
Soil model development from CPT, SPT, and borehole data, translated into dynamic soil parameters appropriate for the chosen wave equation model
Signal matching and back-analysis after installation, comparing predicted and measured behaviour to refine the soil model and improve future predictions
Sensitivity analysis across key parameters including shaft resistance, toe resistance, quake, damping, and hammer efficiency, so you understand which uncertainties drive the result
Iterative learning loops that turn installation monitoring data into an updated model, improving the accuracy of subsequent predictions on the same or similar projects
Structural stress checks during driving, using the continuous model to resolve tension and compression distributions along the pile length

Whether your project involves onshore driven piles, offshore monopiles, or vibro-installed tubular piles, we can help you select the right modelling approach and deliver results you can act on. Contact us to discuss your drivability study or pile testing requirements.