Datasets:

AIMindTeams
/

synthetic-chemical-reactor-5k-sample

Timestamp_min int64 0 5k	Reactor_Temp_C float64 60.5 110	Jacket_Flow_Rate_L_min int64 936 1.06k	Pressure_atm float64 2.22 4.49	Reactant_A_Conc_mol_L float64 0 2.5	Product_B_Conc_mol_L float64 0 2.5	Reactor_Run_ID int64 1 27
0	85	1,015	3.2	2.5	0	1
1	84.5	1,007	3.18	2.49	0.01	1
2	85	1,012	3.22	2.48	0.02	1
3	85.5	1,019	3.16	2.47	0.03	1
4	84	1,002	3.21	2.46	0.04	1
5	85	1,011	3.19	2.45	0.05	1
6	85.5	1,018	3.23	2.44	0.06	1
7	84	999	3.17	2.43	0.07	1
8	85	1,013	3.2	2.42	0.08	1
9	85.5	1,017	3.16	2.41	0.09	1
10	84	1,004	3.22	2.4	0.1	1
11	85	1,015	3.18	2.39	0.11	1
12	85.5	1,016	3.24	2.38	0.12	1
13	84	1,001	3.17	2.37	0.13	1
14	85	1,013	3.2	2.36	0.14	1
15	85.5	1,018	3.19	2.35	0.15	1
16	84	998	3.22	2.34	0.16	1
17	85	1,011	3.18	2.33	0.17	1
18	85.5	1,014	3.24	2.32	0.18	1
19	84	1,006	3.16	2.31	0.19	1
20	85	1,012	3.21	2.3	0.2	1
21	85.5	1,017	3.19	2.29	0.21	1
22	84	1,003	3.23	2.28	0.22	1
23	85	1,016	3.17	2.27	0.23	1
24	85.5	1,019	3.2	2.26	0.24	1
25	84	1,007	3.18	2.25	0.25	1
26	85	1,014	3.22	2.24	0.26	1
27	85.5	1,018	3.19	2.23	0.27	1
28	84	1,002	3.24	2.22	0.28	1
29	85	1,011	3.16	2.21	0.29	1
30	85.5	1,017	3.21	2.2	0.3	1
31	84	1,005	3.18	2.19	0.31	1
32	85	1,013	3.23	2.18	0.32	1
33	85.5	1,016	3.17	2.17	0.33	1
34	84	1,009	3.2	2.16	0.34	1
35	85	1,012	3.16	2.15	0.35	1
36	85.5	1,015	3.22	2.14	0.36	1
37	84	1,007	3.19	2.13	0.37	1
38	85	1,014	3.24	2.12	0.38	1
39	85.5	1,018	3.17	2.11	0.39	1
40	84	1,003	3.21	2.1	0.4	1
41	85	1,011	3.18	2.09	0.41	1
42	85.5	1,017	3.25	2.08	0.42	1
43	84	1,006	3.16	2.07	0.43	1
44	85	1,013	3.22	2.06	0.44	1
45	85.5	1,019	3.19	2.05	0.45	1
46	84	1,002	3.23	2.04	0.46	1
47	85	1,015	3.17	2.03	0.47	1
48	85.5	1,018	3.2	2.02	0.48	1
49	84	1,008	3.16	2.01	0.49	1
50	85	1,015	3.2	2.01	0.49	1
51	85.5	1,010	3.18	2	0.5	1
52	86	1,018	3.19	1.99	0.51	1
53	85.5	1,007	3.21	1.98	0.52	1
54	86	1,012	3.17	1.97	0.53	1
55	85	1,019	3.16	1.96	0.54	1
56	85.5	1,009	3.22	1.95	0.55	1
57	86	1,014	3.19	1.94	0.56	1
58	85.5	1,011	3.2	1.93	0.57	1
59	86	1,006	3.18	1.92	0.58	1
60	85	1,013	3.17	1.91	0.59	1
61	85.5	1,008	3.21	1.9	0.6	1
62	86	1,015	3.19	1.89	0.61	1
63	85.5	1,010	3.2	1.88	0.62	1
64	86	1,007	3.18	1.87	0.63	1
65	85	1,012	3.17	1.86	0.64	1
66	85.5	1,014	3.21	1.85	0.65	1
67	86	1,009	3.19	1.84	0.66	1
68	85.5	1,016	3.2	1.83	0.67	1
69	86	1,008	3.18	1.82	0.68	1
70	85	1,013	3.17	1.81	0.69	1
71	85.5	1,011	3.21	1.8	0.7	1
72	86	1,006	3.19	1.79	0.71	1
73	85.5	1,015	3.2	1.78	0.72	1
74	86	1,007	3.18	1.77	0.73	1
75	85	1,012	3.17	1.76	0.74	1
76	85.5	1,014	3.21	1.75	0.75	1
77	86	1,009	3.19	1.74	0.76	1
78	85.5	1,016	3.2	1.73	0.77	1
79	86	1,008	3.18	1.72	0.78	1
80	85	1,013	3.17	1.71	0.79	1
81	85.5	1,011	3.21	1.7	0.8	1
82	86	1,006	3.19	1.69	0.81	1
83	85.5	1,015	3.2	1.68	0.82	1
84	86	1,007	3.18	1.67	0.83	1
85	85	1,012	3.17	1.66	0.84	1
86	85.5	1,014	3.21	1.65	0.85	1
87	86	1,009	3.19	1.64	0.86	1
88	85.5	1,016	3.2	1.63	0.87	1
89	86	1,008	3.18	1.62	0.88	1
90	85	1,013	3.17	1.61	0.89	1
91	85.5	1,011	3.21	1.6	0.9	1
92	86	1,006	3.19	1.59	0.91	1
93	85.5	1,015	3.2	1.58	0.92	1
94	86	1,007	3.18	1.57	0.93	1
95	85	1,012	3.17	1.56	0.94	1
96	85.5	1,014	3.21	1.55	0.95	1
97	86	1,009	3.19	1.54	0.96	1
98	85.5	1,016	3.2	1.53	0.97	1
99	86	1,008	3.18	1.52	0.98	1

End of preview. Expand in Data Studio

RL-Ready Synthetic Chemical Batch Reactor Dataset

🚨 Download the full 50,000-row dataset featuring 269 complete reactor life-cycles here: https://aimindteam.gumroad.com/l/reactor-dataset 🚨

Overview

This dataset provides a mathematically consistent, high-fidelity simulation of an industrial liquid-phase exothermic batch reactor. It contains multivariate time-series operational logs designed specifically for training Reinforcement Learning (RL) agents, testing Predictive Maintenance (PdM) algorithms, and validating anomaly detection architectures.

Due to the proprietary nature of real-world chemical manufacturing, clean operational data featuring cascading equipment failures is rarely available. This dataset fills that gap by simulating normal baseline operations alongside critical edge-case anomalies, bounded by strict thermodynamic, mass-balance, and kinetic constraints.

Data Dictionary

The dataset is structured as a continuous time-series log with the following schema:

Feature	Data Type	Description
Timestamp_min	Integer	Sequential time step of the reactor operation (minutes).
Reactor_Temp_C	Float	Internal reactor temperature (°C).
Jacket_Flow_Rate_L_min	Float	Volumetric flow rate of the coolant through the reactor jacket (L/min).
Pressure_atm	Float	Internal vessel pressure (atm).
Reactant_A_Conc_mol_L	Float	Molar concentration of the primary reactant (mol/L).
Product_B_Conc_mol_L	Float	Target Variable: Molar concentration of the finished product (mol/L).
Reactor_Run_ID	Integer	Unique identifier for each discrete batch lifecycle (Episode ID for RL).

Simulated Anomalies

To provide realistic failure states for RL training and anomaly detection, the dataset includes dynamically injected physical faults that trigger non-linear system responses:

Cooling System Failure (Exothermic Runaway): A simulated pump failure or jacket fouling represented by a sudden drop or freeze in Jacket_Flow_Rate_L_min, leading to uncommanded spikes in Reactor_Temp_C and Pressure_atm.
Pressure Spikes: Sudden thermodynamic deviations requiring rapid adjustments to jacket flow to prevent thermal runaway.
Suboptimal Yields (Kinetics Shift / Mixing Loss): Unseen degradation in internal mixing efficiency resulting in Reactant_A_Conc_mol_L stagnating and failing to convert optimally into Product_B_Conc_mol_L, despite nominal thermal states.

Recommended Use Cases

Reinforcement Learning: Train agents on discrete episodes (Reactor_Run_ID) to optimize coolant control (Jacket_Flow_Rate_L_min) to maximize Product_B_Conc_mol_L while strictly bounding Reactor_Temp_C.
Predictive Maintenance (PdM): Build early-warning classification models to detect cooling jacket fouling before critical thermal/pressure spikes occur.
Process Optimization: Test metaheuristic algorithms for ideal flow-rate adjustments to compress batch completion times.

License & Usage

Dataset (reactor_sample_5k.csv): Licensed under CC BY-NC 4.0. Free for academic and non-commercial research. Commercial use or deployment in revenue-generating models requires purchasing the full 50,000-row dataset.
Code (Reactor_EDA_and_PoV.ipynb): Licensed under the MIT License. You are free to use, modify, and distribute the analytical code without restriction.

Downloads last month: 4