2. Sled test fundamentals

A MADYMO sled reconstruction is only as trustworthy as the physical test it is built from. Before a single multibody or FE model is assembled, the analyst must understand how the sled pulse was generated, which boundary conditions constrained the occupant environment, what the instrumentation actually measured, and how the acquisition chain shaped the recorded signals. This section establishes those fundamentals and connects each one to the decisions made later in the reconstruction workflow.

Pulse Generation and the Deceleration Corridor

A sled test reproduces the crash deceleration of a vehicle body without destroying a full vehicle. The sled carriage, carrying seats, restraints, and instrumented occupants, is accelerated or decelerated by a propulsion system, and the resulting acceleration–time history at the sled structure is the pulse. Common generation methods include:

• Servo-hydraulic catapult: programmable velocity profile, excellent repeatability, widely used for frontal and side pulses.
• Bungee or pneumatic launch: high acceleration capability, less precise pulse shaping.
• Deceleration sled with crushable or hydraulic absorbers: the sled is accelerated to a target velocity and then decelerated through a programmable energy-absorbing device.

The pulse is specified as a corridor: an upper and lower acceleration bound over time, plus a required velocity change Δv. A typical frontal corridor might require a peak of approximately 35–40 g between 40 and 70 ms with Δv near 50 km/h, but the exact values come from the vehicle crash pulse being reproduced. The analyst must verify that the measured sled pulse lies inside the corridor and that the integrated velocity matches the target Δv. Integration is performed on the filtered acceleration signal:

v(t) = ∫ a(t) dt ,  Δv = v(t_end) − v(0)
s(t) = ∫ v(t) dt  (sled displacement, used for boundary checks)

Two practical cautions matter for reconstruction. First, the pulse used in MADYMO must be the sled structure pulse, not the occupant chest or seat rail acceleration; mixing them introduces a systematic error. Second, the pulse must be filtered consistently with the test data processing (commonly a low-pass filter per SAE J211, e.g. CFC 60 for sled acceleration). Applying a different filter in the model than in the test invalidates the correlation.

Boundary Conditions

Boundary conditions define how the occupant environment is constrained and how loads leave the system. In a sled reconstruction they include:

Each boundary element must be represented in MADYMO with the same coordinate convention used in the test. Establish the origin and axis directions first — for example, +X forward, +Y to the occupant's left, +Z upward, with the origin at the sled reference point — and keep that convention through the entire model. If the test report uses a different sign convention for acceleration (some facilities report deceleration as positive), convert explicitly and document the conversion.

Instrumentation and Data Acquisition

The instrumentation set determines what can be correlated. A standard frontal sled test typically includes:

• Sled accelerometers (triaxial) on the carriage and on the seat rail.
• Occupant accelerometers: head CG triaxial, chest triaxial, pelvis triaxial.
• Load cells: belt load at shoulder and lap, seat belt anchor loads, steering column axial load.
• Displacement / rotation sensors: belt payout, knee slider travel, head rotation via angular rate sensors.
• High-speed video: at least one lateral and one frontal view, with a timing mark synchronized to the data acquisition.

All channels pass through signal conditioning (amplifier, anti-alias filter) and are sampled by a data acquisition system. Sampling rate and filtering are not arbitrary: SAE J211 defines channel classes, and the filter class must match the measurement. For example, occupant head acceleration is typically processed at CFC 1000, while sled acceleration may use CFC 60. The reconstruction must apply the same class to the simulated signal before comparison.

Two acquisition details deserve attention during correlation. First, time zero must be defined consistently — usually the first forward motion of the sled or a specified acceleration threshold. A time offset of even 5 ms shifts every peak and destroys the correlation. Second, channel polarity and units must be verified against the test report; a reversed belt load channel will produce a physically impossible negative tension and should be caught before modeling begins.

From Test Data to Reconstruction Inputs

Once the test data are validated, they are converted into model inputs. The sled pulse is applied as a prescribed acceleration or velocity boundary to the sled reference node. Belt load channels become target curves for the restraint model. Occupant accelerations become correlation targets, not model inputs. The distinction is important: feeding a measured chest acceleration into the model would make the correlation meaningless.

A practical validation sequence before running the reconstruction:

1. Integrate the filtered sled acceleration and confirm Δv within tolerance of the test target (commonly ±1 km/h).
2. Check that the pulse lies inside the corridor at all times, not only at the peak.
3. Verify time zero alignment between pulse, belt loads, and video.
4. Confirm channel polarity: belt loads positive in tension, occupant acceleration positive in the defined forward direction.
5. Confirm the model coordinate origin matches the sled reference point used in the test report.

With a validated pulse, documented boundary conditions, and a consistent instrumentation chain, the MADYMO model has a defensible physical basis. The next stage — building the occupant and restraint model — depends directly on the quality of these inputs, and correlation errors traced back to this stage are far more common than errors in the contact definitions themselves.