Skip to content
TheChatArchiveThe Chat Archive
DeepSeekdeepseek-r1-0528· DeepSeekHugging Face LMArena Human Preference 140K

How much takes analysis of ship collision in Ansys LS-DYNA using rigid body for striking…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-3-5-sonnet-20241022’s answer.
  1. User

    How much takes analysis of ship collision in Ansys LS-DYNA using rigid body for striking vessel and model containing less than 10000 elements of struck vessel on Intel core i9 14900 with 96 Gb of memory and ssd hard drive?

  2. DeepSeek Assistant

    Based on your specifications (rigid striker, <10,000 elements, i9-14900K, 96GB RAM, SSD), the entire LS-DYNA analysis (setup, solution, termination) will likely take between 1 and 3 minutes, often closer to 1 minute.

    Here's the breakdown and reasoning:

    1. Model Size is Extremely Small: <10,000 elements is tiny for LS-DYNA, especially on modern hardware. Ship collision models typically require millions of elements for meaningful results. This model will fit entirely in the CPU's cache and RAM effortlessly.
    2. Rigid Striker: Defining the striking vessel as rigid significantly reduces computational cost. LS-DYNA only tracks its motion (6 degrees of freedom) and calculates contact forces, avoiding the expensive computation of internal element stresses and deformations.
    3. Powerful Hardware:
      • CPU (i9-14900K): 24 cores (8P+16E) with very high clock speeds (up to 6.0 GHz Turbo). Even if LS-DYNA only scales moderately well to 24 cores (which it often does for smaller models before communication overhead dominates), this CPU has immense single-core speed for the serial parts and plenty of cores for parallelization.
      • RAM (96GB): Vastly overkill for this model size. The model data will load almost instantly from the SSD into RAM.
      • Storage (SSD): Ensures very fast read times for the input deck and write times for results (though the result files will also be small).

    Factors Influencing the Estimate:

    • Simulation Duration: The physical time simulated (TERMINATION_TIME) is the biggest factor within this small model context. Simulating 0.1 seconds will be faster than simulating 5.0 seconds.
    • Contact Definition: Complex contact (e.g., AUTOMATIC_SURFACE_TO_SURFACE with refinement) takes slightly more setup than simple contact, but the overhead is negligible at this scale.
    • Output Requests: Writing detailed state data (e.g., GLSTAT, MATSUM, D3PLOT, D3THDT) at very high frequency (every 100 steps) adds some I/O time. Writing less frequently or less data speeds it up slightly.
    • Solver Settings: Using drelax=1 or mass scaling (DT2MS) artificially increases the stable time step, reducing the number of cycles needed. This can significantly speed up small models like this.
    • Initialization: Reading the input deck, allocating memory, setting up contacts/constraints takes a fixed time (seconds). For a tiny model, this becomes a noticeable portion of the total runtime.

    Why "1-3 minutes" is a Safe Estimate:

    • The computational workload per cycle is minimal.
    • The number of cycles needed is low (due to the small size and potential for mass scaling/large time steps).
    • The hardware can process those cycles extremely quickly.
    • Initialization and termination I/O are fast thanks to the SSD and small file sizes.

    Important Caveat:

    • Physical Realism: While this setup will run very fast, a model with <10,000 elements is almost certainly too coarse to produce physically meaningful results for an actual ship collision analysis. It cannot capture the complex plastic deformation, buckling, tearing, and energy absorption mechanisms crucial in such events. Rigid bodies also don't absorb energy through deformation. This setup is only suitable for: Very basic proof-of-concept contact checks, extremely simplified educational examples, or testing solver settings/logic - not for engineering analysis of a real collision scenario. Realistic analyses require hundreds of thousands to millions of elements and deformable structures for both vessels.

    In summary: Expect the job to finish in roughly 1-3 minutes on your powerful hardware due to the extreme simplification (rigid, tiny mesh). However, recognize that these simplifications render the results physically unreliable for a real ship collision study.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.