4. Test data sources and formats

Every injury-limit evaluation ultimately rests on a set of time-series channels recorded during a crash test. Before any Excel pass/fail logic can be trusted, the analyst must understand where those channels come from, how they are stored, how they were sampled and filtered, and what each channel name actually means. This section covers the practical data-handling layer that sits between the test laboratory and the evaluation spreadsheet.

Where the Channels Come From

In a full-scale crash test, the primary measurement source is the anthropomorphic test device (ATD, or "dummy") together with vehicle-mounted sensors. A Hybrid III 50th-percentile male dummy, for example, carries a defined set of transducers: head accelerometers (three orthogonal axes at the head CG), upper and lower neck load cells, chest accelerometers and a chest deflection potentiometer, lumbar spine load cells, and femur/iliac load cells. Each transducer produces one or more analog channels that are digitized by the on-board data acquisition system (DAS) and stored in the test's raw data file.

Alongside the dummy channels, the test typically records vehicle accelerometers (B-pillar, sill, tunnel), seat belt load cells, seat track displacement, and high-speed video time codes. For injury evaluation, the dummy channels are the primary interest, but vehicle channels are often needed to confirm event timing and to align the data with the video.

File Formats: ISO-MME and DIAdem

Two formats dominate European and international crash-test data exchange:

• ISO-MME (ISO 13499) is a directory-based format. A test is stored as a folder containing a binary data file (often .bin or .dat) plus a set of ASCII description files: .txt channel lists, .chn channel definitions, and a .mme master file. The description files carry the channel name, unit, sampling rate, filter class, and sensor location. This separation of data and metadata is what makes ISO-MME portable across labs.
• DIAdem (National Instruments) uses .dat or .tdm/.tdms files with an accompanying channel header. DIAdem files are common when data has already been processed in the lab's analysis environment; the channel properties (unit, description, filter) are embedded in the file header rather than in separate text files.

In practice, the analyst receives either a raw ISO-MME folder or an exported DIAdem file. The first task is always to inventory the channels: confirm that every required dummy channel is present, check units, and verify the sampling rate and filter class recorded in the metadata. A channel that exists but was sampled at the wrong rate or filtered with the wrong class will silently corrupt the injury calculation.

Sampling Rates

Crash-test DAS units typically sample at rates from 10 kHz to 20 kHz per channel. The choice matters because injury criteria involve integration and differentiation of the signal. Head injury criterion (HIC), for instance, integrates the resultant head acceleration over a 36 ms window; a low sampling rate introduces integration error, and a rate that is not an integer multiple of the analysis rate complicates resampling. SAE J211/1 recommends a minimum sampling rate consistent with the channel's filter class, and most labs standardize on 10 kHz or 20 kHz for dummy accelerometers and load cells.

When channels arrive at different rates (for example, a 10 kHz accelerometer and a 1 kHz displacement channel), the analyst must resample to a common rate before combining them. Resampling should be done with an anti-aliasing filter, not by simple decimation.

Filtering per SAE J211

SAE J211/1 defines channel filter classes (CFC) that specify a low-pass Butterworth filter with a defined cutoff frequency. The class is chosen according to the physical quantity being measured:

The filter must be applied in the forward direction only (zero-phase filtering is not used for J211 compliance, because the standard specifies a causal filter). Applying the wrong CFC — for example, evaluating HIC on a CFC 60 signal — will under-predict the peak acceleration and produce a falsely passing result. The filter class is recorded in the ISO-MME channel description; the analyst should verify it rather than assume it.

Channel Naming Conventions

Channel names in crash-test data follow a structured convention that encodes sensor location, measurement type, and axis. A typical ISO-MME name looks like H3_HEAD_ACC_X or H3_NECK_FX. The prefix identifies the dummy (H3 = Hybrid III), the middle segment the body region and transducer, and the suffix the axis or force component. DIAdem exports often use a similar but not identical scheme, so a mapping table between the lab's naming and the evaluation spreadsheet's expected names is essential.

Axis conventions must be fixed before any calculation. In the SAE J211 vehicle coordinate system, +X is forward, +Y is to the right, and +Z is downward. Dummy accelerometer signs follow the same convention, but some labs invert the Z axis for head accelerometers. The analyst must confirm the sign convention from the channel description and keep it consistent across all channels used in a single criterion.

Practical Verification Before Evaluation

A short, repeatable check sequence prevents most data-driven errors. First, open the ISO-MME .chn file and confirm that each required channel is present with the expected unit and CFC. Second, plot the raw signal and check that the event window (typically 0–200 ms) contains a plausible pulse and no clipping or dropouts. Third, apply the J211 filter and compare the filtered peak against the raw peak; a large reduction indicates the raw signal contained high-frequency noise that the filter correctly removed, while an unchanged peak suggests the filter may not have been applied. Fourth, confirm the axis sign by checking that the head acceleration during the forward impact is positive in +X.

Once the channels are verified, filtered, resampled, and mapped to the spreadsheet's expected names, the data is ready for the injury-limit calculations. The next stage of the workflow takes these conditioned time series and applies the HIC, Nij, chest deflection, and femur force formulas that form the core of the pass/fail evaluation.