Run a Single Simulation

First, we will run a single OpenFAST simulation using the 5MW_OC4Semi_WSt_WavesWN case. This should be straightforward as all the necessary input files are already prepared.

Code Overview

The code required to run an OpenFAST simulation using the Inductiva API is always the same. We just need to adapt it for this use case, as shown below.

import inductiva

# Allocate Cloud machine on Google Cloud Platform
cloud_machine = inductiva.resources.MachineGroup(
   provider="GCP",
   machine_type="c2d-highcpu-2",
   spot=True)

# Initialize OpenFAST simulator
openfast = inductiva.simulators.OpenFAST(
   version="4.0.2")

# Run simulation
task = openfast.run(
   input_dir="input_files",
   commands=[
       "openfast 5MW_OC4Semi_WSt_WavesWN/"
       "5MW_OC4Semi_WSt_WavesWN.fst"],
   on=cloud_machine)

# Wait for the simulation to finish and download the results
task.wait()
cloud_machine.terminate()

task.download_outputs()

task.print_summary()

When the simulation is complete, we terminate the machine, download the results and print a summary of the simulation as follows.

inductiva tasks info hk8fmb4dif05c5vf5saurocxp


Task status: Success


Timeline:
   Waiting for Input         at 20/02, 14:24:28      6.487 s
   In Queue                  at 20/02, 14:24:35      19.415 s
   Preparing to Compute      at 20/02, 14:24:54      1.842 s
   In Progress               at 20/02, 14:24:56      34.265 s
       └> 34.123 s        openfast 5MW_OC4Semi_WSt_WavesWN/5MW_OC4Semi_WSt_WavesWN.fst
   Finalizing                at 20/02, 14:25:30      1.085 s
   Success                   at 20/02, 14:25:31


Data:
   Size of zipped output:    16.20 MB
   Size of unzipped output:  32.71 MB
   Number of output files:   83

Total estimated cost (US$): 0.01013 US$
    Estimated computation cost (US$): 0.00013 US$
    Task orchestration fee (US$): 0.010 US$

Note: A per-run orchestration fee (0.010 US$) applies to tasks run from 01 Dec 2025, in addition to the computation costs.
Learn more about costs at: https://inductiva.ai/guides/basics/how-much-does-it-cost

Performance and Cost Analysis

Given that OpenFAST does not benefit from multiple CPU cores, we chose the c2d-highcpu-2 virtual machine (VM) with 2 virtual CPUs (equivalent to 1 physical core). This is one of the cheapest options on Google Cloud, costing just US$0.0081 per hour in spot mode.

To demonstrate that OpenFAST does not scale with the number of cores, we also ran the same simulation on a number of better machines. For a detailed breakdown, check out our Benchmarks section.

Here are the results:

MachinevCPUsExecution timeEstimated Cost (USD)
c2d-highcpu-2234.1s0.00013
c2d-highcpu-4435.0s0.00031
c2d-highcpu-8830.4s0.00034
c2d-highcpu-161630.9s0.00065
c2d-highcpu-323230.2s0.0012

The execution time remains almost the same on all machines, regardless of the number of virtual CPUs.

Additionally, running the simulation on the c2d-highcpu-2 VM proves to be extremely cost-efficient, with a total cost of only 0.00013 US$.

In the next part of this tutorial, we'll take things to the next level by running dozens of OpenFAST simulations in parallel on Inductiva, demonstrating the true power of cloud-based scalability. Stay tuned!