As discussed in my previous post, DoE allows us (hopefully) to extract maximum insight from minimal runs. This post summarizes a two-round DoE strategy used to optimize final titer in a microbial fermentation process using two 12 x 250 mL ambr bioreactor setups across two weeks of experimentation. All data is fabricated and hypothetical.
Round 1: Exploring DO and pH We began with a 3×4 full factorial design to explore the effects of initial dissolved oxygen (DO) and pH on final titer. Twelve bioreactors were used in a single run:
- DO levels: 2%, 10%, 25%
- pH levels: 6.5, 6.75, 6.95, 7.2
This resulted in 12 unique condition combinations, each run once. Results indicated a clear trend: the highest titers occurred at the lowest DO (2%) and a medium pH (6.75). These results were used to lock in DO and pH for a second round of experimentation.
Round 1 Summary Table:
| DO (%) | pH | Titer (g/L) |
|---|---|---|
| 2 | 6.50 | 9.1 |
| 2 | 6.75 | 10.0 |
| 2 | 6.95 | 9.3 |
| 2 | 7.20 | 8.9 |
| 10 | 6.50 | 8.5 |
| 10 | 6.75 | 9.2 |
| 10 | 6.95 | 8.7 |
| 10 | 7.20 | 8.3 |
| 25 | 6.50 | 7.6 |
| 25 | 6.75 | 7.9 |
| 25 | 6.95 | 7.2 |
| 25 | 7.20 | 6.9 |
Round 1 Visualization: Round 1 Heatmap
Round 2: Zooming in on Temperature With DO (2%) and pH (6.75) fixed, the next round focused on testing temperature as the primary variable. Four temperatures were tested: 30°C, 33°C, 36°C, and 39°C. Each condition was run in triplicate, using all 12 bioreactors.
Key Findings:
- 33°C produced the highest titers, averaging just above 10 g/L.
- ANOVA analysis showed temperature had a statistically significant impact on final titer (p < 0.001).
- A strong downward trend was observed at 36°C and 39°C, indicating stress or reduced metabolic efficiency.
Round 2 Sample Results:
| Temperature (°C) | Titer Replicates (g/L) |
|---|---|
| 30 | 8.65, 8.46, 8.69 |
| 33 | 10.66, 10.13, 10.13 |
| 36 | 9.47, 9.23, 8.86 |
| 39 | 7.66, 7.36, 7.36 |
Round 2 Visualizations:
Boxplot:
Bar Chart with Error:
Why This Works: The Layered Approach By using a layered DoE strategy, we were able to:
- Efficiently explore broad conditions (Round 1)
- Focus deeply on a single variable (Round 2)
- Maintain statistical significance while minimizing reactor usage
This iterative method ensures that each round of experimentation builds on the last, avoiding redundancy and accelerating discovery.
Final Thoughts The combination of factorial design, clear hypothesis framing, and targeted follow-up gives us the tools to make smarter, faster decisions.
Keywords: Design of Experiments, fermentation optimization, bioreactor, titer improvement, pH, DO, temperature, JMP, factorial design.
Leave a Reply