Penicillium: a Python implementation of the IndPenSim penicillin fermentation simulator

IndPenSim for Python: validated, pip-installable, fast. Penicillium is a maintained implementation of IndPenSim, the industrial-scale (100 m³) fed-batch penicillin fermentation simulator of Goldrick et al. (2015, 2019), the benchmark process for bioprocess monitoring, soft sensors, fault detection and control.

• Installs anywhere. Python 3.10 or later; tested on Linux, macOS and Windows with Python 3.10 to 3.14; no compiler, no MATLAB.
• Follows IndPenSim. Checked line by line against IndPenSim V2.01; 14 of 15 batches of V2.01 (run in GNU Octave) replayed within 0.1 % on yield. Against the published 100-batch data set, no significant difference detected in the yield distributions of recipe-driven, operator-controlled and faulty batches (KS p = 0.31-0.93; the tests have low power) (validation).
• Fast. 0.6 s per 230 h batch; about 380 batches per minute in parallel on an Intel Core i7-7700HQ (benchmarks).
• Ready for ML and control. One-call campaigns with labels and batch-wise splits, Gymnasium environments, Minari datasets and a differentiable JAX model.
• Reproducible. Random inputs bit-identical and trajectories within about 1e-4 on Linux, macOS and Windows.
• Honest about its limits. Fixed fault windows, true values in the sensor-fault channels and the other known limitations are listed on the first page of the documentation (limitations).
• Open licence. MIT, with the parts derived from IndPenSim also under IndPenSim's BSD 3-Clause licence; documented provenance (licensing).

Status: alpha (0.1.0a1). The API may still change.

Documentation: https://penicillium.readthedocs.io

Install

pip install "penicillium[fast]"        # with the Numba backend (recommended)

Other extras: gym, minari, jax, dcs, parquet, notebooks, docs, dev (installation).

Example

from penicillium import Simulator

result = Simulator().run(seed=1)  # one 230 h batch, 1150 samples of 12 min
print(result.total_yield)  # harvested penicillin, kg
df = result.to_pandas()  # 55 channels, indexed by time in hours

Learn more

All of these are on the documentation site, https://penicillium.readthedocs.io.

From a clone of the repository, run the tests: pytest -m "not slow" (seconds) or pytest (adds full batches, ~4 min).

Citation

If you use Penicillium, cite IndPenSim and this package (CITATION.cff; see citing):

• S. Goldrick, A. Ştefan, D. Lovett, G. Montague, B. Lennox. The development of an industrial-scale fed-batch fermentation simulation. Journal of Biotechnology 193 (2015) 70-82. doi:10.1016/j.jbiotec.2014.10.029
• S. Goldrick, C. A. Duran-Villalobos, K. Jankauskas, D. Lovett, S. S. Farid, B. Lennox. Modern day monitoring and control challenges outlined on an industrial-scale benchmark fermentation process. Computers and Chemical Engineering 130 (2019) 106471. doi:10.1016/j.compchemeng.2019.05.037

Licence

Penicillium is released under the MIT License. Its process model, controllers, recipes, fault profiles and reference Raman spectrum derive from IndPenSim, Copyright (c) 2016, Stephen Goldrick, distributed under the BSD 3-Clause License; that notice applies to those parts too. It is not endorsed by IndPenSim's authors or by University College London. Full texts are in LICENSE and NOTICE.