Open Source Bioreactor: Commissioning
The first run, once the reactor is together as Building it leaves it, and the numbers worth sending back from it.
Set the gas rate before anything else. 0.1-0.5 vvm on 8.17 L is 0.817-4.09 L/min, which is 16 % to 82 % of the registered 0-5 L/min meter - both ends on scale, which is what that range buys. The pump is seven times oversized, so the entire band sits in the first part of the needle valve’s travel - that is why the valve is a 3 deg needle and not the 20 deg one.
Watch the DO trace on the first run. On a shaft-drive build with the loop or the ring, where the
upper impeller’s flow number is known, head() reports where the DO probe’s face sits against the
sparger (DO probe in the gas:; of the transcripts, build__jar_10L_220x305 carries it), and warns
when the face sits above the sparger and over its radius. A galvanic probe reads high with a bubble
on the membrane and low with nothing moving past, so a probe over the bubbles can be wrong in both
directions. A characteristically spiky trace means the probe is in the plume. Where the bubbles go
is a bench question, and the model has two levers: the port’s lean (do_probe_port_tilt_max, a
ceiling) and the sparger’s height (sparge_cap_drop for the loop, sparge_ring_gap_fraction for
the ring). The lean leans outward, so leaning less moves the face inward. On jar_10L the loop is
inboard and the lean already at its 4.5 deg ceiling, so leaning less only brings the face nearer the
loop: 28.7 mm off its radius at 4.5 deg, 18.2 at 0. With the ring, on jar_6p5gal_305x470, the face
sits 0.59 mm outboard of the ring’s radius at the ceiling, crosses it near 4.25 deg, and clears the
warning only at 0, 9.89 mm inboard. The height moves the gas up or down under the face. The echo
gives the face’s position after either change.
Know the replace-the-filter signal. The sterile filter can rise about 50 % - from 14.1 kPa to roughly 21.1 - before 0.5 vvm becomes unreachable with the valve wide open. So valve fully open and still short of the rate is the signal, and it is not a pump problem.
What to record
Section titled “What to record”The model is waiting on four measurements. If you build one of these, these are the numbers worth sending back:
- The sterile filter’s Delta P at the set flow, from a water manometer across it. It is the largest term in the gas budget and the only one that is extrapolated.
- The jar’s rim, calipered - the land width on top of the glass, the wall, and the mouth. Three registered numbers disagree on two vessels and a caliper says which is wrong.
- The baffle’s lean. Print the lid’s baffle port and one plate segment, seat them, and measure
the tilt at a known distance below the lid. The model keeps 1 mm running clearance after a
stacked worst case of lean, 2.44 mm at
jar_10L’s lower impeller, and cannot tell how much of that lean is real. On the reference build the lock bore caps the plate first, so the lean costs no width there; on a build where the impeller caps it, whatever the measurement takes off is plate width back. - Whether the joint holds. Nobody has yet tightened twelve nuts a third of a turn onto a glass jar with a printed flange in between. How much it relaxes, and over what time, is the number the build most needs.
