Open Source Bioreactor: References
Every external source this design rests on, what it is used for, and where in the repo that use lives. Kept so a number in the CAD can be traced back to the thing that justifies it, and so nobody has to reconstruct the reasoning later.
Verification status is recorded per source. It says how far each source was actually checked, since a claim relayed through a summary can be wrong:
- read - the full text was retrieved and the cited passage confirmed
- abstract - only the abstract was available; the claim is from it
- unread - cited from a secondary source, not obtained
Grades: [PR] peer-reviewed - [TH] thesis - [PAT] patent - [TX] textbook - [TP] trade press - [VN] vendor.
Every literature source below is also in references.bib - 64 entries, every one
carrying its grade and verification status in a note field, so the whole set imports into a
reference manager in one go rather than one DOI at a time. Vendor pages, part numbers and software libraries
are deliberately not in it: they are cited inline on the registry row or file that uses them.
Impeller and tank geometry
Section titled “Impeller and tank geometry”Fitschen, J., Maly, M., Rosseburg, A., Wutz, J., Wucherpfennig, T. & Schluter, M. (2019).
“Influence of Spacing of Multiple Impellers on Power Input in an Industrial-Scale Aerated Stirred
Tank Reactor.” Chemie Ingenieur Technik 91(11):1794-1801. doi:10.1002/cite.201900121 [PR] -
read
D/T 0.3-0.5 and ~0.3 for radial impellers; impeller spacing 1.0-2.0 d; the up-to-35 % power
penalty for spacing too close. Their own rig ran d/D = 0.33, a radial ratio, so the spacing rule is
carried from the literature they review rather than measured on an axial impeller.
And the literature it is carried from is Davis. Fitschen’s ref for both the 1.0-2.0 d band and
his eq. (5) impeller-count inequality is his [37] = Davis (2010) - the thesis graded unreliable
below, whose own two refs contain neither claim. So this entry does not re-source the spacing band
away from that thesis, which is what it was being used for; it relays it. His [33] Uhl & Gray (1966)
and [35] Hudcova, Machon & Nienow (1989) carry the 35 % power penalty, not the band.
Fitschen eq. (5) is still what stirred_tank_impeller_count_bounds() implements, because it is
the only stated form of the rule anyone here has found and the arithmetic follows from the band
regardless - but it is reported, never asserted, and there is no primary behind it.
-> scad/utils/stirred_tank.scad, scad/head.scad impeller guideline block
McCabe, W.L., Smith, J.C. & Harriott, P. (1993). Unit Operations of Chemical Engineering, 5th
ed. McGraw-Hill. ISBN 0-07-044844-2 [TX] - read
The standard turbine (ch. 9, pp. 242-243, Fig. 9.7 after Rushton, Costich & Everett, Chem. Eng.
Prog. 46:395, 467, 1950): D/T 1/3, H/T 1, baffles T/12 and four of them, the impeller T/3 off the
floor, blades D/5 high and D/4 long, 6 or 8 of them. With several impellers the bottom one sits
about one impeller diameter off the bottom (p. 241), and two turbines at least one impeller diameter
apart draw about 1.9 times one (p. 253). The reference to this book came from the microalgae
researcher advising the project. It gives no disc diameter, and its own power number for the disc
turbine, 5.75 at T/10 baffles (Table 9.3), is not the one the model uses
(Agitation).
-> scad/custom/impellers.scad (the Rushton’s blades and tank layout), Baseline
Rutherford, K., Lee, K.C., Mahmoudi, S.M.S. & Yianneskis, M. (1996). “Hydrodynamic
characteristics of dual Rushton impeller stirred vessels.” AIChE Journal 42(2):332-346.
doi:10.1002/aic.690420204 [PR] - abstract
Three dual-Rushton layouts at H = T: parallel flow (lower 0.25T off the floor, the pair 0.5T apart),
merging flow (equal thirds), diverging flow (lower 0.15T). The reference build runs the merging
flow’s equal thirds, which is also McCabe’s T/3 off the floor and one impeller diameter apart. Read
through Li et al. 2007’s Fig. 1 and the abstract; the thresholds between the patterns are not
claimed here.
-> scad/custom/impellers.scad
Li, Z., Gao, Z., Smith, J.M. & Thorpe, R.B. (2007). “Large Eddy Simulation of Flow Fields in
Vessels Stirred by Dual Rushton Impeller Agitators.” Journal of Chemical Engineering of Japan
40(8):684-691. doi:10.1252/jcej.40.684 [PR] - read
Fig. 1 relays Rutherford et al.’s three layouts with their clearances.
-> scad/custom/impellers.scad
Paul, E.L., Atiemo-Obeng, V.A. & Kresta, S.M. (eds.) (2004). Handbook of Industrial Mixing: Science and Practice. Wiley-Interscience. [TX] - unread Relayed by a literature search and not checked against the text: baffles T/12 to T/10 (ch. 6 p. 350), D and C each T/4 to T/2 (Fig. 11-1 p. 592), one impeller up to Z/T 1.0 for gas dispersion and two to 1.8 (Table 6-6 p. 372). -> Baseline
Nienow, A.W. (2006). “Reactor engineering in large scale animal cell culture.” Cytotechnology
50(1-3):9-33. doi:10.1007/s10616-006-9005-8 [PR] - read
Design guideline (a): axial hydrofoils at 0.4-0.5 of the vessel diameter, dual, clearance between
them 0.33-0.5 T, sparger below the lower impeller, baffles required. Also the dimensional objection
to tip speed, the honest Kolmogorov statement, the four regions of bubble damage, and the estimate
of peak dissipation from power over impeller swept volume.
-> scad/utils/stirred_tank.scad, scad/head.scad
Nienow, A.W. (2021). “The Impact of Fluid Dynamic Stress in Stirred Bioreactors - The Scale of
the Biological Entity: A Personal View.” Chemie Ingenieur Technik 93(1-2):17-30.
doi:10.1002/cite.202000176 [PR] - read
Restates the tip-speed objection fourteen years on; turbulence at Re > ~2e4; “high” and “low shear”
impeller labels described as largely disproven.
-> scad/utils/stirred_tank.scad
Rotondi, M. et al. (2021). Biotechnology Letters 43:1103-1116.
doi:10.1007/s10529-021-03076-3 [PR] - read
As impeller diameter approaches 50 % of vessel diameter, axial flow loses its strong axial motion -
the upper bound on D/T.
-> scad/utils/stirred_tank.scad
Fort, I., Jirout, T., Sperling, R., Jambere, S. & Rieger, F. (2002). “Study of Pumping Capacity
of Pitched Blade Impellers.” Acta Polytechnica 42(4). doi:10.14311/380 [PR] - read
Power number against blade angle and blade count, and the Medek correlations
Po scales as (sin alpha)^2.077 and N_Qp as (sin alpha)^0.468 (valid alpha 15-60 deg, Re > 1e4, four baffles) - power
rises with angle about 4.4x faster than pumping does. Blade-count series 3/4/6 -> 0.79/0.99/1.34.
Also gives the blade width. Both impeller families are
Czech Standards and both are drawn at h/D = 0.2 - the simple PBT is CVS 691020 (nB 3 at alpha 24 deg,
35 deg, 45 deg; nB 6 at alpha 45 deg), the folded-blade impeller CVS 691010 (nB 3, s/D 1.5, alpha 67 deg, beta 25 deg, gamma 48 deg).
Measured in a T = 400 mm vessel with four wall baffles at H = T, D/T 0.36. Two caveats for
transferring it: H/T = 1.0 there against 1.09 here, which is the one Medek departure this
vessel cannot remove; and the folded blade is a three-angle fold, not a flat plate at a pitch, so
its measured Po does not transfer to a plate drawn at 45 deg.
-> scad/custom/impellers.scad impeller_folded_axial_3/4/6 and impeller_pbt_45_4 width_ratio,
scad/head.scad impeller_n_fins
Medek, J. & Fort, I. - the correlations this model actually computes with, reached through
Fort et al. (2002) above rather than obtained. [PR] - unread
stirred_tank_medek_power_number() and ..._flow_number() implement Po scaling as (sin alpha)^2.077 and
N_Qp as (sin alpha)^0.468 as Fort et al. state them; the primary papers have not been read. Crossref
gives two candidates in Coll. Czech. Chem. Commun. - “Pumping effect of impellers with flat
inclined blades,” 44:3077-3089 (1979), doi:10.1135/cccc19793077,
and Fort, Medek & Placek, “Hydraulic characteristics of paddle impellers with flat inclined blades,”
40:3443-3458 (1975), doi:10.1135/cccc19753443 - but WHICH
of them Fort et al. drew from has not been checked, so neither is claimed here.
-> scad/utils/stirred_tank.scad stirred_tank_medek_*
Jirout, T. & Rieger, F. “Impeller design for mixing of suspensions.” CTU Prague -
https://users.fs.cvut.cz/tomas.jirout/vyuka/p2_hmp/chep_vyuka.pdf [TX] - read
Reproduces the above and adds the folded-blade series, including Np = 0.99 +/- 0.04 for a 4-blade
folded axial impeller - the closest measured analogue to this project’s blade.
-> scad/custom/impellers.scad impeller_folded_axial_4
Zhou, G., Shi, L. & Yu, P. (2003). “CFD Study of Mixing Process in Rushton Turbine Stirred
Tanks.” 3rd Int. Conf. on CFD in the Minerals and Process Industries, CSIRO, Melbourne -
https://www.cfd.com.au/cfd_conf03/papers/002Guo.pdf [PR] - read
The classic standard tank configuration: H = T, D = T/3, C = T/3, four baffles at T/10. Also gives
the reference Rushton: six blades with width and height both D/4.
-> scad/custom/impellers.scad impeller_rushton_6
Grenville, R., Giacomelli, J., Padron, G. & Brown, D. (2017). “Impeller Performance in Stirred
Tanks.” Chemical Engineering, August 2017, pp. 46-54 -
https://framatomebhr.com/Portals/0/PDF/Publications/Impeller-Perfomance-in-Stirred-Tanks.pdf
[TP] - read
The peak-dissipation correlation epsilon_max = 1.04*x*Po^(3/4)*N^3*D^2 (+/-15 %) and the peak-to-mean form
0.82*(x/Po^(1/4))*(T/D)^3. Trade press, not peer-reviewed - the weakest-graded source carrying
real weight in this design. The impeller constant x - Rushton 12, pitched blade 16, hydrofoil 17 -
is a property of the shape, so it is carried per type.
-> scad/custom/impellers.scad x column, scad/utils/stirred_tank.scad
Lonza Biologics, US10883076B2 - https://patents.google.com/patent/US10883076B2/en [PAT] - read Axial impellers at D/T 0.35-0.55, preferred 0.40-0.48, “most preferred” 0.44-0.46. Self-contradictory on its own upper bound; superseded as the D/T citation by Nienow 2006 and Rotondi 2021.
Davis, R.Z. (2010). “Design and Scale-Up of Production Scale Stirred Tank Fermentors.” MS thesis, Mechanical and Aerospace Engineering, Utah State University - https://digitalcommons.usu.edu/etd/537 [TH] - read Spacing 1.0-2.0 d with the bottom impeller 1.0 d off the floor, and power falling to about 80 % of the properly-spaced value below 1.0 d; top impeller >= 1.5 d below the liquid surface; baffles 0.08-0.10 T, four on 90 deg centres. Davis relays these rather than measuring them, from his refs [15] and [17]. Both primaries have been read, and four of the five relayed claims are not in either of them.
| Davis’s claim | In Oldshue [15]? | In Xing [17]? |
|---|---|---|
| spacing 1.0-2.0 d between impellers | no | no |
| bottom impeller 1.0 d off the floor | partly - see below | no |
| power falls to ~80 % below 1.0 d | no | no |
| top impeller >= 1.5 d below the surface | no | no |
| baffles 0.08-0.10 T, four on 90 deg centres | partly - four at 1/12 T | no |
- [15] = Oldshue (1997), chapter 5 of the handbook - retrieved and read, see its own entry below. The one thing that matches is p. 192: “If the impeller can be placed one to two impeller diameters off bottom…” - an off-bottom clearance for fluidfoil impellers, stated as a conditional trade-off. Davis appears to have read that as an impeller-to-impeller spacing rule; the numbers 1-2 d are the same, the quantity is not. Nothing in the chapter gives an impeller separation, an 80 % penalty, or a submergence depth.
- [17] = Xing, Z., Kenty, B.M., Li, Z.J. & Lee, S.S. (2009), Biotechnol Bioeng 103(4), doi:10.1002/bit.22287. Retrieved and read. A mixing-time, kLa and CO2-removal scale-up study on 5,000 L CHO bioreactors that contains no impeller-spacing guideline at all - “impeller distance” and “off-bottom clearance” appear only as nomenclature symbols for its own vessel. (Davis also cites it Kenty-first; Xing is the first author.)
Treat this entry as unreliable for numbers. Cite Oldshue directly for the off-bottom clearance and the baffles.
So the spacing band, the 1.0 d off-bottom clearance, the 80 % penalty and the baffle geometry all trace to a handbook nobody here has opened, relayed through a thesis. Treat them as leads, not as design targets; the read against the thesis says only that Davis says them.
Oldshue, J.Y. (1997). “Agitation,” chapter 5 of Vogel, H.C. & Todaro, C.L. (eds.),
Fermentation and Biochemical Engineering Handbook: Principles, Process Design, and Equipment,
2nd ed., Noyes Publications, Westwood NJ, 1997, pp. 181-241. ISBN 0-8155-1407-7. [TX] - read
The primary behind the numbers Davis (2010) relays. What it actually says:
- Baffles, p. 202. Four baffles, “each 1/12 the tank diameter in width” - a single value, 0.0833 T, not the 0.08-0.10 band Davis reports. And explicitly: “Either 3, 6 or 8 baffles can be used if preferred. The general principle is to use the same total projected area as exists with four baffles, each 1/12 the tank diameter in width.” The constraint is total projected area, not the count; how the lid-plate builds fall short of it is in Agitation.
- Off-bottom clearance, p. 192. “If the impeller can be placed one to two impeller diameters
off bottom, which means that mixing is not provided at low levels during draw off, these
impellers offer an excellent flow pattern as well as considerable economies in shaft design.”
Read the grammar: “if … can … offer” makes this an allowance for fluidfoil impellers with
a stated cost, not a rule to meet.
head()reports a departure from it rather than warning. - Coverage over the impeller, p. 191-192 - the sentence immediately before the band, and the one that decides this reactor’s clearance. “However, there is a tendency for these impellers to short-circuit the fluid to a relatively low distance above the impeller. Very careful consideration of the coverage over the impeller is important.” He gives no number. On the pitched pair at D/T 0.45 the two requirements do not overlap: the band needs C >= 94.5 mm and half a diameter of coverage allows C <= 69.2, because two impellers span 154.5 mm of a 241 mm column.
- Axial impellers in a gassed system, p. 228. “the upward flow of gas tends to negate the
downward action of the pumping capacity of the axial flow turbine. A radial flow turbine must
have three times more power than the power in the gas stream for the mixer power level to be
fully effective. On the other hand, the axial flow impeller must have eight to ten times more
power than in the gas stream for it to establish the axial flow pattern.”
head()evaluates the radial rule for the two Rushtons at each drive speed (gas against the lower impeller:); what it gives is in Agitation. At the rated 320 rpm the alternative 94.5 mm pitched pair (D/T 0.45) holds its axial pattern to 0.70 vvm against the 0.5 vvm design. - Sparger, p. 214. “A sparge ring about 80 % of the impeller diameter is more effective than an open pipe beneath the impeller or sparge rings larger than the impeller… the desired entry point for the gas is where it can pass initially through the high shear zone around the impeller.” Two experimental studies find the opposite - Birch & Ahmed 1997 and Rewatkar & Joshi 1993, both above - and both recommend rings larger than the impeller, which is what the model follows. What survives is the second half: gas should enter where the impeller immediately works it, which agrees with Nienow 2006.
- Gas dispersion. The lower impeller does most of the gas-dispersing work, and a common three-impeller power split is 40 % lower / 30 % each upper.
- kLa by gassing out, p. 226. “an unsteady state reaeration test in which the tank is stripped of oxygen; air is started with the mixer running and the dissolved oxygen level increase is monitored until the tank is saturated” - the method this project needs, since Van’t Riet’s correlation does not reach its P/V. He warns the endpoint sits between top and bottom saturation values, so the driving force used to reduce it is not the log-mean one.
Not in it, despite being attributed to it: any impeller-to-impeller spacing rule, the 80 % power penalty below 1.0 d, and any top-impeller submergence depth. -> Agitation, sparger
Kaiser, S.C., Werner, S., Jossen, V., Kraume, M. & Eibl, D. (2017). “Development of a method
for reliable power input measurements in conventional and single-use stirred bioreactors at
laboratory scale.” Engineering in Life Sciences 17(5):500-511.
doi:10.1002/elsc.201600096 [PR] - read
The power-number definition Np = P/(rho N^3 d^5) and measured Rushton Np 4.17 +/- 0.14. Cited in the SCAD
as “Kaiser 2016”, which is the online publication date; the issue is 2017.
-> scad/custom/impellers.scad impeller_rushton_6
Sparger geometry and location
Section titled “Sparger geometry and location”Together these overturn the 80 %-of-impeller ring Oldshue gives on p. 214 - see that entry.
Birch, D. & Ahmed, N. (1997). “The Influence of Sparger Design and Location on Gas Dispersion
in Stirred Vessels.” Chemical Engineering Research and Design (Trans IChemE Part A) 75(A5):487-496.
doi:10.1205/026387697523994 [PR] - read
From the abstract: “Significant performance improvements, in terms of improved power draw and
delayed onset of flooding on aeration, are achieved through the use of ‘larger than impeller’
ring spargers, positioned within the discharge stream from the impeller. There is little or no
penalty in terms of the gas holdup generated.” They tested a ring at 1.4 D, which is not
arbitrary - the annulus from R out to 1.41 R encloses the same volume the impeller sweeps, a
physical rationale Oldshue’s 80 % has never had. Gas belongs in the impeller’s discharge stream,
so above an up-pumping impeller and below a down-pumping one.
-> scad/utils/stirred_tank.scad stirred_tank_sparge_ring_band() lower bound
Rewatkar, V.B. & Joshi, J.B. (1993). “Role of Sparger Design on Gas Dispersion in Mechanically
Agitated Gas-Liquid Contactors.” The Canadian Journal of Chemical Engineering 71(2):278-291.
doi:10.1002/cjce.5450710215 [PR] - read
“The use of a large ring is recommended”, with the critical speed for gas dispersion lowest at
a ring twice the impeller diameter. Also: “Hole size and number of holes on the ring sparger have negligible effect when the sparger
is located near the impeller. However, these variables become important when the sparger is
located away from the impeller.” They used 2, 3 and 6 mm holes. Scale caveat: vessels of 0.57,
1.0 and 1.5 m diameter against this project’s 0.21 m, and a pitched-blade downflow turbine,
where this reactor’s lower impeller is a radial Rushton.
-> scad/utils/stirred_tank.scad stirred_tank_sparge_ring_band() upper bound
Palfy, T.G., Torok, L., Kalicz, P. & Gribovszki, Z. (2021). “3D printed receptacle with
diffuser membrane for manipulating pressurized air and water.” HardwareX 9:e00192.
doi:10.1016/j.ohx.2021.e00192 [PR] - read
The one printed air-permeable membrane found with a measured point: PETG from a 0.4 mm nozzle, three
0.2 mm layers on a 15 % honeycomb, walls at 115 % extrusion to seal and the membrane at 95 % by
editing the G-code. Table 2: 14.7 L/h at 0.2 bar through a 46.8 x 46.8 mm membrane (2190 mm^2),
with 50 entry points, all within 838 mm^2 of it, and 1-3 mm bubbles; the active area shrank 20-25 %
over about forty hours and twenty freezing cycles. One flow, and the paper does not say where the
0.2 bar was read. So it is a membrane with fifty uncontrolled holes rather than an air stone, and a
single point scales only by assumption.
-> scad/custom/sparge_puck.scad palfy_*(), borrowed, scaled linearly
Blade twist - measured in the literature, but out of this blade’s range
Section titled “Blade twist - measured in the literature, but out of this blade’s range”Twist is characterised, with a consistent direction of effect. What is missing is coverage of the twisted paddle’s range.
Both papers define blade twist as the hub angle minus the tip angle. On that definition the
twisted paddle is a 30 deg twist - 83 deg at the hub falling to 53 deg at the tip
(Agitation) - which
is 1.5x the largest twist either paper tested, at hub angles 23 deg steeper than either explored.
impeller_twist_ang and impeller_height therefore remain uncharacterised design parameters,
documented by derivation in scad/head.scad rather than by citation.
- Patwardhan, A.W. & Joshi, J.B. (1999). Ind Eng Chem Res. doi:10.1021/ie980772s [PR] - read - section 3.4.3 tests 10 deg and 20 deg twist on 30 deg, 45 deg and 60 deg blades: “For a particular blade angle, an increase in the twist decreases the values of N_P and N_QP. However, the values of N_QS are affected to a much smaller extent. As a result of these, the mixing time decreases with an increase in the blade twist.”
- Kumaresan, T. & Joshi, J.B. (2006). Chem Eng J 115:173-193. doi:10.1016/j.cej.2005.10.002 [PR] - read - section 3.1.3.3, 10 deg twist on a six-blade 30 deg pitched turbine lowers the power number, cuts secondary flow number 3 % and averaged shear rate 1 %. Small in magnitude, same direction as Patwardhan.
Consequence for the model. Two independent studies agree that twist reduces the power number, so the Po = 0.99 of an untwisted 4-blade folded axial impeller is most likely an over-estimate for the twisted paddle, and anything derived from it is conservative in the safe direction. Neither paper reaches 30 deg twist at 83 deg hub angle, so this is a direction, not a correction factor.
- Lightnin / General Signal patents - https://patents.google.com/patent/US5158434A/en, https://patents.google.com/patent/US4468130A/en [PAT] - read - the only published twist numbers found (30-45 deg twist, 18-30 deg tip chord angle). They describe a cambered foil with a tangential chord; this project’s blade is a thick radial fin, so they do not transfer.
- Helical-ribbon and screw-impeller work is creeping-flow, Re < 20, four orders from this reactor’s duty, and does not apply: Ameur et al. 2018, doi:10.3390/chemengineering2020026 [PR] - read - its simulations run at Re = 10.
Cell damage, shear and agitation
Section titled “Cell damage, shear and agitation”Molina Grima, E., Acien Fernandez, F.G., Garcia Camacho, F. & Chisti, Y. (1999).
“Photobioreactors: light regime, mass transfer, and scaleup.” Journal of Biotechnology
70(1-3):231-247. doi:10.1016/S0168-1656(99)00078-4
[PR] - read (archived reprint Asterio.pdf)
Source of the “frequency of switch” criterion Agitation quotes - that a photobioreactor
is judged by how often a cell crosses between lit and dark zones, not by a blend-time multiple.
Confirmed on p. 234: average irradiance “considers only the total length of the dark and the light
periods, not the frequency of switch.”
-> Agitation mixing-criterion paragraph
Mazzuca Sobczuk, T., Garcia Camacho, F., Molina Grima, E. & Chisti, Y. (2006). “Effects of agitation on the microalgae Phaeodactylum tricornutum and Porphyridium cruentum.” Bioprocess and Biosystems Engineering 28(4):243-250. doi:10.1007/s00449-005-0030-3 [PR] - read The central mechanism: “Mechanical agitation was not the direct cause of cell damage. Damage occurred because of the rupture of small gas bubbles at the surface of the culture.” Tip speeds
1.56 m/s (P. tricornutum) and 2.45-2.89 m/s (P. cruentum). Pluronic F-68 protective.
Leupold, M., Hindersin, S., Gust, G., Kerner, M. & Hanelt, D. (2013). “Influence of mixing and shear stress on Chlorella vulgaris, Scenedesmus obliquus, and Chlamydomonas reinhardtii.” J Appl Phycol 25:485-495. doi:10.1007/s10811-012-9882-5 [PR] - read Optimum tip speed 1.26 m/s = 0.45 Pa (+48 % growth for C. vulgaris, +71 % for S. obliquus); at 2.03 m/s = 0.9 Pa photosynthetic activity falls back to the unstirred control; at 5.89 m/s it is 7-8 % below control. C. reinhardtii does not benefit (-2.3 % PA at the optimum).
Read the apparatus before transferring the tip speed. This is not a stirred tank with an
impeller. It is a Gust (1989) microcosm - 1.75 L of culture in PMMA, “mixing was created by the
tip speed of a spinning plate with skirt” sitting 5 cm above the bottom, with metered fluid
recirculated through the centre axis. Their u_tip = r*omega is that plate’s rim speed.
Three consequences for how this project uses the numbers:
- The shear is generated in a plate-to-wall gap, not an impeller discharge. The paper puts the main force at “a surface of 644 cm^2 at the wall and spinning plate”, as a Reynolds stress in the water head between the two. An impeller of the same tip speed produces a discharge jet, which is a different mechanism. Equal tip speed does not imply equal shear between the two devices, and the paper does not claim it does.
- The Pa figures are the bottom stress, which the authors call the secondary force. Calibrated friction velocity “occurs at 314 cm^2 of the bottom of the microcosm and is the secondary, minor shear force in present study”. So the 0.45 and 0.9 Pa quoted above are not the dominant stress in their own experiment.
- Direct calibration stops at 250 rpm. Rates above that “are calculated by means of an empirical model based on hydrodynamics of a boundary layer theory (Schlichting 1968)”.
None of this makes the measurement wrong, and it remains the only Chlorella-specific agitation figure this project has. What it means is that treating 1.26 and 2.03 m/s as impeller tip speeds is an assumption this project makes, not a result the paper reports - and it is the assumption that sets the operating band and, through it, D/T. The physically transferable quantity the authors themselves emphasise is friction velocity; converting that to an impeller would require estimating the shear our own impeller produces, which has not been done. -> Agitation
Hu, W., Berdugo, C. & Chalmers, J.J. (2011). “The potential of hydrodynamic damage to animal cells of industrial relevance.” Cytotechnology 63(5):445-460. doi:10.1007/s10616-011-9368-3 [PR] - read With no gas-liquid interface, cells are unaffected at 100-450 rpm, 600 rpm in a full vessel, and 1500 rpm for hybridomas. Bubble rupture 1e7-1e9 W/m^3. A review, so those three results are restated from Zhang & Thomas 1993 and Kunas & Papoutsakis 1990.
Ma, N., Koelling, K.W. & Chalmers, J.J. (2002). Biotechnol Bioeng 80(4):428-437. doi:10.1002/bit.10387 [PR] - read Bubble-rupture dissipation orders of magnitude above impeller dissipation: rupture reaches 1.66e4-4e5 kW/m3 against no detected damage below 1e4 kW/m3, and the peak falls two orders as bubble diameter grows 1.7 to 6.32 mm.
Barbosa, M.J., Albrecht, M. & Wijffels, R.H. (2003). Biotechnol Bioeng 83(1):112-120. doi:10.1002/bit.10657 [PR] - read Damage originates at the sparger, during bubble formation, rather than at bursting: “Superficial gas velocity alone cannot be used to estimate cell damage in sparged microalgal cultures, which means that bubble bursting is not the only factor, and might not even be the most important factor, leading to cell death.” Gas entrance velocity is named as the parameter to watch, damage is strain-dependent, and the cell wall is confirmed protective. No critical velocity is established. Barbosa’s own runs were 0.4-5.4 m/s; the larger figures quoted in it (13.3 m/s lab, 47.8 m/s pilot) are two comparison reactors from Camacho et al. 2001, and the paper closes “more work remains to be done to clarify the influence of this parameter”.
Michels, M.H.A. et al. (2010). Bioprocess Biosyst Eng 33:921-927, doi:10.1007/s00449-010-0415-9; and Michels, M.H.A.; van der Goot, A.J.; Vermue, M.H.; Wijffels, R.H. (2016). “Cultivation of shear stress sensitive and tolerant microalgal species in a tubular photobioreactor equipped with a centrifugal pump.” Journal of Applied Phycology 28(1):53-62. doi:10.1007/s10811-015-0559-8 [PR] - read Shear thresholds in Pa for four marine species; damage is instantaneous rather than cumulative; the rigid-cell-wall rule; measured shear in real photobioreactors (circulation tube 0.57 Pa, pump pressure side 1.82 Pa, centrifugal impeller tip 26 Pa). The 2016 paper prints J Appl Phycol (2016) 28:53-62 on the page and 20 March 2015 as the online date. Context only - it tests four marine aquaculture species and not Chlorella. Useful for scale: severe damage to I. galbana, S. costatum and C. muelleri between 1.2 and 5.4 Pa, while T. suecica tolerates 80 Pa. Shear tolerance spans more than an order of magnitude between species, which is why a number borrowed from one alga says little about another.
Wang, C. & Lan, C.Q. (2018). Biotechnology Advances 36(4):986-1002. doi:10.1016/j.biotechadv.2018.03.001 [PR] - read Taxonomic shear-tolerance ranking: greens > cyanobacteria > haptophytes > reds > diatoms > dinoflagellates. Chlorella is a rigid-walled green alga, the most tolerant class.
Varley, J. & Birch, J. (1999). Cytotechnology 29(3):177-205. doi:10.1023/A:1008008021481 [PR] - read The provenance of the widely-quoted 1.5 m/s tip-speed limit - attributed to Kioukia et al. 1992 for human melanoma cells in 5 % serum, then generalised, in a paper that also reports 6 m/s causing no measurable harm to a hybridoma line. Also the correct rule that large slow agitators beat small fast ones at equal power input. Cited here as a caution, not as a design input.
Brindley Alias, C. et al. (2004). Biotechnol Bioeng 87(6):723-733. doi:10.1002/bit.20179 [PR] - read - damage above 30-80 s^-1 in a pump-driven tubular photobioreactor.
Contreras, A., Garcia, F., Molina, E. & Merchuk, J.C. (1998). Biotechnol Bioeng 60(3):317-325 doi:10.1002/(SICI)1097-0290(19981105)60:3<317::AID-BIT7>3.0.CO;2-K [PR] - read - growth optimum at ~7000 s^-1 in a concentric tube airlift. These two are often presented as contradictory. They are not - different equipment, different shear mechanism.
Godoy-Silva, R. et al. (2009). Biotechnol Bioeng 103(6):1103-1117. doi:10.1002/bit.22339 [PR] - read - CHO resistant to the highest EDR tested, 6.4e6 W/m^3, with most product quality attributes unaffected; the glycosylation shift begins at 6.0e4 W/m^3, two orders lower.
Walls, P.L.L. et al. (2017). Sci Rep 7:15102. doi:10.1038/s41598-017-14531-5 [PR] - read - bubble-rupture energy simulation; repeated exposure lowers the lethal threshold.
Karimi Alavijeh, M., Lee, K. & Gras, S.L. (2024). Eng Life Sci 24(7):e2400023. doi:10.1002/elsc.202400023 [PR] - read - measured Kolmogorov length 78 um against a <20 um cell, so the criterion is not binding at these scales.
Uyar, B., Ali, M.D. & Uyar, G.E.O. (2024). Bioprocess Biosyst Eng 47(2):195-209. doi:10.1007/s00449-023-02952-8 [PR] - read The only peer-reviewed power budget found for a microalgal stirred tank: aeration 29.4-49.1 W/m^3 against stirring 0.65-1.30 W/m^3.
Wileman, A., Ozkan, A. & Berberoglu, H. (2012). Bioresour Technol 104:432-439. doi:10.1016/j.biortech.2011.11.027 [PR] - read Microalgal broth rheology: Newtonian with <30 % viscosity rise below 20 kg/m^3. Covers Chlorella, Nannochloris and P. tricornutum - not Arthrospira.
Daub et al. (2014). J Biol Eng 8:17, PMC4099098. [PR] - unread Reported as measuring that larger impellers give a lower ratio of maximum to specific dissipation at equal power per volume: the measured case for Agitation’s large-and-slow.
Podgorska (2019). Entropy 21:340. [PR] - unread Reported as finding that a low-Np hydrofoil made smaller droplets than a Rushton at equal P/V: a case for section 4’s caution that low shear is not reliably gentle.
Seals and glands
Section titled “Seals and glands”Apple Rubber, Seal Types and Gland Design Tables, Section 4 -
https://www.applerubber.com/src/pdf/section4-seal-types-and-gland-design-tables.pdf [VN] -
read
Table A static-seal bands: 19-33 % axial squeeze, 14-23 % radial; groove width ratios by cord; fill
never above 85 %, hard limit 90 %. Table B for the groove’s outer diameter under internal pressure.
-> bayonet-port-scad’s utils/oring_gland.scad, scad/head.scad lid seal block
Bolted joint
Section titled “Bolted joint”ASME (2019). ASME Boiler and Pressure Vessel Code, Section VIII, Division 1: Rules for
Construction of Pressure Vessels. ASME BPVC.VIII.1-2019, Mandatory Appendix 2, “Rules for Bolted
Flange Connections With Ring Type Gaskets,” Table 2-5.1 p.399. [TX] - read
Gasket factor m and minimum design seating stress y, for
“elastomers without fabric or high percent of mineral fiber”: below 75A Shore, m = 0.50 and
y = 0 psi; 75A or higher, m = 1.00 and y = 200 psi (1.4 MPa). Self-energizing types, o-rings among
them, are m = 0 and y = 0. All four figures the model encodes were confirmed against the table
rather than the summary they came from.
Two things the table says. y is zero for this build’s 60A sheet,
so there is no seating-stress minimum to meet and the model is right to compute none. And the
table’s own note reads: the values “are suggested only and are not mandatory” - so this is
guidance the design follows, not a requirement it satisfies, and a reviewer should be told which.
-> scad/utils/bolt_pattern.scad, scad/head.scad head_gasket_factor()
Codesign Engineering - “Flange Design per Appendix 2” (LinkedIn article, re-hosted) - https://codesignengg.com/wp-content/uploads/2025/05/LinkedIn-Article-Flange-Design-per-Appendix-2.pdf [TP] - read The working reference the m values were actually taken from. A trade-press summary of the entry above, re-hosted on a consultancy’s WordPress: it is where the number came from, so it is cited as such rather than dressed up as the standard.
Baffles
Section titled “Baffles”Baffle width and number calculation -
https://myengineeringtools.com/references/pages/baffle_width_and_number_calculation.html [VN] -
read -> bayonet-port-scad’s custom/bayonet_port.scad bayonet_baffle_port
Dynamix Agitators. “Baffled by Baffles.” Technical note, uploaded 2015.
https://www.dynamixinc.com/wp-content/uploads/dlm_uploads/2015/12/Dynamix-Baffled-by-Baffles.pdf
[VN] - read
A baffle stands a sixth of its width off the wall: J/6, which at T/12 is T/72, the wall cage’s gap. The Handbook
of Industrial Mixing’s T/50 to T/72 reached this project through a trade article and is unread.
-> scad/custom/baffle_cage.scad, Agitation
Sato, E., Ochi, Y., Horiguchi, H., Takenaka, K., Wu, J. & Parthasarathy, R. (2021). “Effect of
Baffle Clearance on Scale Deposition in an Agitated Vessel.” ACS Omega 6(37):24070-24074.
doi:10.1021/acsomega.1c03503 [PR] - read
Flow accelerating behind a baffle keeps that region from going stagnant, which is why partial
baffles standing inboard are not purely a loss. -> scad/head.scad baffle block
Baffle inserts for jars, the prior art for the wall cage. [VN] [PAT] - read
The Eppendorf BioFlo 115 uses a 316L split ring that springs against the glass (manual M1369-0050
section 4.7.4 and section 4.8.4). US4494878 has separate PTFE blades with a top ring, a snap-on bottom ring and
wall standoff tabs. Corning EP3807394, now withdrawn, is a squeeze-and-release four-baffle insert.
None goes in through a mouth narrower than the bore, and no open-hardware design that does was
found. -> scad/custom/baffle_cage.scad, Decisions
Note on status: partial/inboard baffles are characterised in the literature, not recommended. The relevant papers (Chem Eng Res Des S0263876207730952, Chem Eng Sci S0009250906007950) are paywalled and unread. The lid-plate builds’ three inboard baffles rest on an unresolved citation.
Unbaffled vessels and eccentric agitation
Section titled “Unbaffled vessels and eccentric agitation”Mouths under about 98 mm cannot hold four baffles beside two dia 16 probes at any port count (Port layout), so the small jars in the family are unbaffled whatever else is decided.
Montante, G.; Bakker, A.; Paglianti, A.; Magelli, F. (2006). “Effect of the shaft eccentricity on the hydrodynamics of unbaffled stirred tanks.” Chemical Engineering Science 61:2807-2814. doi:10.1016/j.ces.2005.09.021 [PR] - read PIV plus RANS CFD in a T = 23.6 cm closed vessel, Rushton turbine D = T/3 at C = T/2, 400 rpm, Re 4.1e4, above this design’s rated Re. The vessel carried a lid deliberately, “in order to avoid uncertainties in the velocity measurements due to air bubbles entrainment”, so this is a closed vessel with no free-surface vortex, like ours. The number that matters: with the shaft centred and no baffles, the measured impeller flow number is 0.25, about 65% below the same impeller in a baffled vessel (p. 2811, agreeing with Brunazzi 2003 by LDA). That is the cost of leaving a vessel unbaffled with a centred shaft, and it is large. Moving the shaft to E = T/4 destroys the double-loop structure and the solid-body swirl, replacing it with two vortices - one inclined about 10 deg from vertical toward the bottom, one about 30 deg toward the top. The paper is a CFD-validation study and reports no mixing times and no flow number for the eccentric case, so it establishes that eccentricity changes the flow, not by how much it improves mixing. It attributes that improvement to Hall et al. (2004), which is read and summarised below - Montante establishes that eccentricity changes the flow, and Hall is what says by how much. Also useful as a modelling caution: k-epsilon predicts unphysical rigid-body rotation in this geometry and steady-state RANS misses the vortices entirely; the Reynolds stress model with a transient sliding mesh was needed.
Galletti, C.; Pintus, S.; Brunazzi, E. (2009). “Effect of shaft eccentricity and impeller blade thickness on the vortices features in an unbaffled vessel.” Chemical Engineering Research and Design 87:391-400. doi:10.1016/j.cherd.2008.11.013 [PR] - unread Two vortices, above and below the impeller, with the upper one dominating and its inclination set by the eccentricity. Notes the vortex oscillates slowly rather than sitting steady, which is a caution rather than a recommendation: it means macro-mixing in an eccentric unbaffled vessel is unsteady.
Scargiali, F.; Busciglio, A.; Grisafi, F.; Brucato, A. (2014). “Mass transfer and hydrodynamic
characteristics of unbaffled stirred bio-reactors: Influence of impeller design.” Biochemical
Engineering Journal 82:41-47. doi:10.1016/j.bej.2013.11.009
[PR] - read
T = 190 mm closed vessel with a mechanical shaft seal, six impellers compared including a PBT at
D = T/3, 100-1300 rpm. Splits the behaviour at a critical speed N_crit where the surface vortex
reaches the impeller and starts ingesting gas.
Sub-critical (no bubbles at all, transfer through the vortex surface): kLa tops out at 1.3e-3 to
3.5e-3 1/s depending on impeller, against the 2.8e-4 1/s minimum for animal cell culture - so about
ten times what such a culture needs, with no bubble bursting anywhere in the vessel. The PBT is
recommended by name: “consistently more efficient in terms of power requirements for achieving a
given kLa” and “may be recommended as a best choice for growing shear sensitive cells”. Its critical
point is N_crit 650 rpm, P/V 126 W/m3, Np 0.52, kLa 2.34e-3 1/s (Table 1).
Super-critical: unbaffled becomes comparable to standard sparged baffled reactors, but only above
about 500 W/m3 - this design’s mean dissipation across the drive’s band
(Agitation) sits below that crossover.
Both regimes here are unsparged. For a vessel that is both unbaffled and sparged, as ours would
be, the paper points to Cabaret et al. (2008), Chem Eng Sci 63:1636-1647, who measured kLa of
0.01-0.06 1/s - an order above the vortex route and in the normal range for a stirred bioreactor.
Note this is the opposite strategy to the one above: eccentric mounting suppresses the vortex,
this uses it. They are two different vessels, not one recommendation.
Busciglio, A.; Scargiali, F.; Grisafi, F.; Brucato, A. (2016). “Oscillation dynamics of free vortex surface in uncovered unbaffled stirred vessels.” Chemical Engineering Journal 285:477-486. doi:10.1016/j.cej.2015.10.015 [PR] - unread
Labik, L.; Petricek, R.; Moucha, T.; Brucato, A.; Caputo, G.; Grisafi, F. (2018). “Scale-up and viscosity effects on gas-liquid mass transfer rates in unbaffled stirred tanks.” Chemical Engineering Research and Design 132:584-592. doi:10.1016/j.cherd.2018.01.051 [PR] - unread
Hall, J.F.; Barigou, M.; Simmons, M.J.H.; Stitt, E.H. (2004). “Mixing in Unbaffled High-Throughput Experimentation Reactors.” Industrial & Engineering Chemistry Research 43:4149-4158. doi:10.1021/ie049872q [PR] - read The source Montante was pointing at: six-bladed 45 deg pitched blade turbines, up-pumping, in vessels of T = 60 and 88 mm - the scale of the small jars in this family. All configurations compared at a constant 168 W/m3, which is what makes the comparison fair. Table 2, mixing time to 95% homogeneity, on the T = 60 vessel: baffled 1.98 s, unbaffled centred 2.80 s, unbaffled eccentric 1.78 s. Off-centre is 36% faster than unbaffled centred and, at equal power per volume, indistinguishable from baffled - the paper puts it at “no discernible difference (<=0.1 s)”. Bulk-fluid energy dissipation improved about 200%. The eccentric position tested was e = 0.2 T, and it was the only one tested.
T is the tank’s internal diameter, which this project’s use of the paper turns on. Two places in
it say so independently: the Nomenclature gives “T = tank diameter (m)” alongside “R = tank radius”
and “Cw = impeller clearance from vessel sidewall”, and Table 1 registers Cw/T as 0.5 for both the
baffled and the unbaffled centred runs and 0.7 for the eccentric one - a shaft half a diameter
from the wall is on the axis, so 0.7 - 0.5 is the eccentricity and e = 0.2 T is measured on the
bore. Table 1 also fixes the fourth row: T88E’s 2.21 s is the 88 mm vessel, so the three figures
above are one vessel and are comparable to each other.
Where it departs from this design: Table 1 gives D/T of 0.58 and 0.55 against 0.33 here,
H/T = 1 against 1.09 on jar_10L, and Re of 8237-19280 against about 2.6e4. The scale matches;
the impeller class does not, a pitched blade against this build’s Rushtons, and nor do the diameter
ratio and the fill. The 36% is measured on a fatter axial impeller in a shallower vessel.
Karcz, J.; Cudak, M.; Szoplik, J. (2005). “Stirring of a liquid in a stirred tank with an eccentrically located impeller.” Chemical Engineering Science 60:2369-2380. doi:10.1016/j.ces.2004.11.018 [PR] - read What Hall does not give: the shape of the curve between centred and 0.2 T. Eq. (6), p. 2372, correlates dimensionless mixing time against eccentricity continuously:
Theta = 48.5 + (1/M) * exp[-1.89 * (1/M) * (e/R)] * exp[8.18 * M]Theta = n * t_m, M = +0.32 for an up-pumping propeller and -0.32 for a down-pumping one, and
R = T/2 - so e/R is twice e/T, which is the conversion this project needs and the easiest
thing in the equation to get wrong. Mean relative error +/-10 %.
Its envelope, from Table 1 and the text: unbaffled, T = 0.7 m inner diameter, H = T,
V = 270 dm3, a propeller at D = 0.33 T with pitch S = D and blade width w = 0.2 D,
off-bottom clearance h = 0.33 H, e/R in [0, 0.57] - so e/T up to 0.285, which covers every
value this design can reach - and Re in [2e4, 8e4], where this design sits at 2.6e4. Table 1’s
first row reads “Inner diameter of the tank, T”, which is the second independent confirmation that
T is the bore.
The response is exponential, not linear, so most of the benefit arrives early: half of it by e/T = 0.05, and 26% of the centred mixing time gone by e/T = 0.07. Both are eq. (6) evaluated, not read off a figure.
Cross-checked against Hall at the one point they share, e/T = 0.2: Karcz’s equation predicts 42% faster than centred, Hall measured 36%. Different impellers, tanks 12x apart in size, agreeing within six points - which is why the correlation, not Hall’s single point, is used in Agitation.
The departures are larger than Hall’s. D = 0.33 T, as here, is not one; a three-bladed
propeller, or an HE 3, rather than this build’s Rushtons; H = T against 1.09; and 270 L
against 8.17. Worst of all, M has only two values and this design is neither - eq. (6) offers
up-pumping or down-pumping, the build’s two Rushtons pump radially, and the two branches disagree
about the centred case by a factor of nearly two (Theta 91.3 against 48.3). Every figure quoted
from eq. (6) in this document is the up-pumping branch, which is the one Hall’s up-pumping PBT
corroborates.
Cabaret, F.; Fradette, L.; Tanguy, P.A. (2008). “Gas-liquid mass transfer in unbaffled dual-impeller mixers.” Chemical Engineering Science 63:1636-1647. doi:10.1016/j.ces.2007.11.028 [PR] - abstract The sparged unbaffled case, which is the one this reactor would actually be in. Its literature review states the consensus plainly - “shaft eccentricity is equivalent to baffling” in turbulent single-phase mixing - and adds two things the others do not: power consumption rises with eccentricity, and an off-centred impeller produces smaller bubbles, so the gain is not only in blending. Dual Rushton, as here. The kLa range of 0.01-0.06 1/s quoted above is Scargiali’s characterisation of this paper, not read off it here.
Scarsella, M.; Torzillo, G.; Cicci, A.; Belotti, G.; De Filippis, P.; Bravi, M. (2012). “Mechanical stress tolerance of two microalgae.” Process Biochemistry 47:1603-1611. doi:10.1016/j.procbio.2011.07.002 [PR] - read Settles the species question, and not in the direction assumed here. Its two test organisms are Chlorella vulgaris and Scenedesmus dimorphus, and the conclusion is that “C. vulgaris cells appear to be weaker than S. dimorphus cells”. What damaged it was a centrifugal pump, at 262 W/kg and 18e3 Pa s of accumulated stress per hour of operation. What did not damage it was an air-lift at 31 W/kg and 90-450 Pa s/h. This design’s peak dissipation is 27.1 W/kg at its rated 320 rpm and 61.2 W/kg at no-load 420 rpm - the rated figure sits just under the non-damaging air-lift and roughly 10x under the damaging pump. Read the caution with it: the paper shows the Kolmogorov criterion mispredicted its own results (both rigs had eddies larger than the cells, yet only one caused damage), and that accumulated stress in Pa s was the better descriptor. Nothing in this model computes that.
Ruszkowski, S. (1994). “A rational method for measuring blending performance, and comparison of
different impeller types.” Proc. 8th European Conference on Mixing, IChemE, p. 283. [PR] -
unread, but validated
The blend-time correlation the model reports, t95 = 5.9 T^(2/3) eps^(-1/3) (T/D)^(1/3). Taken from
Hall 2004 eq. (13) rather than from the proceedings, which were not obtained - so the source is
unread, but the equation is not unchecked: it reproduces Hall’s own published table, 1.96 s against
his 1.9 for a 60 mm vessel and 2.59 against 2.6 for an 88 mm one. Fitted on 0.01-10 m3 fully baffled;
this reactor is 0.0082 m3, so the model echoes the extrapolation every render.
Hall prints a second correlation beside it, Cooke’s t90 = 3.3 (1/N) Po^(-1/3) (T/D)^(1/3). That
one is deliberately not encoded - it could not be made to reproduce the figures Hall prints for it,
and a correlation that fails its own published check is not one to build a report on.
-> scad/utils/stirred_tank.scad, scad/head.scad transfer block
Van’t Riet, K. (1979). “Review of Measuring Methods and Results in Nonviscous Gas-Liquid Mass
Transfer in Stirred Vessels.” Industrial & Engineering Chemistry Process Design and Development
18:357-364. doi:10.1021/i260071a001 [PR] - unread
The standard first estimate for kLa, in two forms because coalescence sets the bubble size and so
the interfacial area: 0.026 (P/V)^0.4 us^0.5 coalescing, 0.002 (P/V)^0.7 us^0.2 not. Miracle-Gro
at 0.2 g/L is dilute enough to be the coalescing case, which gives 0.0081 1/s at the rated 320 rpm -
just under the 0.01-0.06 1/s Cabaret measured for sparged unbaffled tanks, which is the only independent
check available here.
Cited second-hand and used outside its range: the correlation is fitted over 500-10000 W/m3 and this
vessel runs 130. Air-water, besides. The model echoes both caveats rather than presenting the number
bare.
-> scad/utils/stirred_tank.scad, scad/head.scad transfer block
The airlift
Section titled “The airlift”The hydrodynamics here were fitted on loops far taller than these jars: Chisti’s circulation data
spans a stated 1.36-8.5 m of reactor height, against 0.17-0.19 m of liquid here. Only Fernandes’ 3.7 L columns
are this size, and they are split cylinders. What carries over is geometry stated as ratios. Chisti’s scans are the author’s own reprints, archived
from his Massey University page (web.archive.org/web/2016id_/http://www.massey.ac.nz/~ychisti/).
Chisti, M.Y. & Moo-Young, M. (1987). “Airlift reactors: characteristics, applications and
design considerations.” Chemical Engineering Communications 60(1-6):195-242.
doi:10.1080/00986448708912017 [PR] - read
(archived reprint Airlift1987.pdf, pp. 202-225)
p. 213: “we recommend that the total cross-sectional area for fluid flow from the downcomer to the
riser, i.e. the area just below the baffles separating the downcomer and the riser in internal loop
devices … should not exceed 1.65 times the downcomer cross-section.” The reason given is a
bubble-free zone under the baffle that grew as the clearance did. Observed in a rectangular vessel
sparged at the sides, the flow the other way round from this draft tube. The same page recommends
a prism or cone under the turn in annulus-sparged draught tubes, which is the B variant’s
concern and not this one’s.
-> head_draft_tube_end_gap(): the gap gives the turn the downcomer’s own area, 1.0 of it
Chisti, M.Y., Halard, B. & Moo-Young, M. (1988). “Liquid circulation in airlift reactors.”
Chemical Engineering Science 43(3):451-457.
doi:10.1016/0009-2509(88)87005-2 [PR] -
read (archived reprint Halard.pdf)
The circulation model, eq. (16), and the loss where the flow turns under the tube, eq. (18):
K_B = 11.402 (A_d/A_b)^0.789, “over a A_d/A_b range of 0.2-1.8” (p. 455). Fig. 2 draws A_b
for a draught tube as the band under it. At equal area K_B is 11.4, mid-fit. The model itself
is not run: its riser holdup, eq. (19), holds “for U_Lr > 0.3 m s^-1”, its 13 reactors were
1.36-8.5 m tall, and a 2-5 mm/s riser gas velocity is far outside any of it. Also p. 455: Hatch’s
draught tube, at a 3.3 cm clearance, where “a 67 % reduction in this area led to a 32.5 % decline
in the riser superficial liquid velocity” - why the gap is sized rather than left to fall where it
may.
-> head_draft_tube_bottom_loss(), the draft tube areas: line head() reports
Chisti, M.Y. (1989). Airlift Bioreactors. Elsevier, London. [TX] - read, chapter 5
only (archived Chapter5b.pdf, pp. 190-211)
p. 194: “the upper baffle clearance in internal-loop airlifts should be kept small - about 0.15 to
0.20 m”, in 3.5-6 m split cylinders. These jars’ top gaps are 11 and 38 mm, well inside it; no
source read states a top clearance scaled to diameter, so the top gap takes the bottom’s rule.
-> head_draft_tube_end_gap()
Sanchez Miron, A., Garcia Camacho, F., Contreras Gomez, A., Molina Grima, E. & Chisti, Y. (2000).
“Bubble-column and airlift photobioreactors for algal culture.” AIChE Journal 46(9):1872-1887.
doi:10.1002/aic.690460915 [PR] - read (archived
reprint AIChE.pdf)
Bubble column, split cylinder and draft-tube airlift, 0.193 m across, about 2 m of liquid, 0.06 m^3,
Phaeodactylum tricornutum: “there was no significant difference in culture performance in the
three bioreactors”, 0.022 h^-1 at 109 W/m^3 (p. 1885; conclusion 7, p. 1886). The draft tube was 0.144 m
inside, 0.091 m off the floor, riser-to-downcomer ratio 1.24 (p. 1875). The loss they calculate
from that geometry, K_B 4.6, is reproduced by eq. (18) with A_b over the tube’s inside diameter -
11.4 x 0.319^0.79 = 4.62 - which is the perimeter head() takes. So the airlift is not chosen
for productivity: this is the paper that says it buys none over a column at equal power. It is
chosen for the defined circulation Agitation argues for.
-> head_draft_tube_end_gap()
Sanchez Miron, A., Ceron Garcia, M.-C., Garcia Camacho, F., Molina Grima, E. & Chisti, Y. (2002).
“Growth and biochemical characterization of microalgal biomass produced in bubble column and airlift
photobioreactors: studies in fed-batch culture.” Enzyme and Microbial Technology 31(7):1015-1023.
doi:10.1016/S0141-0229(02)00229-6 [PR] -
read (archived reprint Asterio Paper.pdf)
The same three vessels. p. 1020: in the column cells from the dark core moved freely to the lit
wall, whereas “the vertical baffle and the draft-tube prevented radial motion of fluid” - so a
draft tube replaces radial mixing with a loop through lit and dark, and an opaque one would make
the core a place cells only pass through in the dark.
-> draft_tube_* in head.scad: print it clear
Fernandes, B.D., Mota, A., Ferreira, A., Dragone, G., Teixeira, J.A. & Vicente, A.A. (2014). “Characterization of split cylinder airlift photobioreactors for efficient microalgae cultivation.” Chemical Engineering Science 117:445-454. doi:10.1016/j.ces.2014.06.043 [PR] - read (repositorium.uminho.pt) p. 446, on concentric tubes: a “large fraction of dark zones inside the PBR, mainly due to the presence of the internal column, which limits light penetration”; their own baffles were transparent PMMA “to allow light penetration”. Ran riser:downcomer 1.0 and 3.0 in 90 mm columns, and 1.0 grew better. -> print the tube clear
Klein, J., Szijjarto, A., Vicente, A.A. & Teixeira, J.A. (2002). “Hydrodynamic considerations in
three-phase internal-loop airlift bioreactors - effect of dual separator and draught tube design.”
29th Conference of the Slovak Society of Chemical Engineering. [PR] - read
(repositorium.uminho.pt)
Draft-tube sparged, 60 L, A_D/A_R 1.2 against 3.97: the wide downcomer ran about a quarter of the
riser’s liquid velocity and needed U_G 0.03 m/s before any bubble was drawn down it - a gentler
loop. These jars’ 0.54 and 1.72 fall inside what Chisti’s circulation data covers, not Klein’s;
no source read gives an optimal A_d/A_r for an internal loop, so the model reports it and does
not target it.
-> the draft tube areas: line
Probe mounting
Section titled “Probe mounting”Yokogawa TNA1505, Best Practice pH Installation and Maintenance (2015) -
https://web-material3.yokogawa.com/TNA1505_+pH+Installation+and+Maintenance+Manual.pdf
[VN] - read
“the sensor should be mounted with the process flow coming towards the sensor, and positioned at
least 15 deg above the horizontal plane to eliminate air bubbles in the pH glass bulb.” A probe hanging
straight down sits 90 deg above horizontal, so vertical clears that by 75 deg. This is why the pH port
leans 0 deg at no cost to the probe - it is not a concession made to fit the vessel.
-> scad/head.scad ph_probe_port_tilt_degrees
Atlas Scientific, lab-grade DO probe datasheet -
https://files.atlas-scientific.com/LG_DO_probe.pdf [VN] - read
Galvanic, and it CONSUMES the oxygen it reads: “a small amount of water movement is necessary to
take accurate readings. Approximately 60 ml/min”. The chart beside it is the part that matters -
stagnant water takes the reading from 90 % to 20 % in thirty seconds, so a still probe does not
read a bit low, it reports the hole it has eaten around itself. Silent on mounting angle.
-> scad/head.scad do_probe_flow_requirement, and stirred_tank_circulation_velocity() for what
the vessel offers against it
-> bayonet-port-scad’s purchased/atlas_probes.scad do_lab_g2
Atlas Scientific, lab-grade pH probe datasheet -
https://files.atlas-scientific.com/pH_probe.pdf [VN] - read
Silent on mounting angle. Its only bubble note is about air inside the probe’s own stem after
shipping, cleared by shaking it down “as done with a clinical thermometer” - not about bubbles on
the sensing face.
-> bayonet-port-scad’s purchased/atlas_probes.scad ph_lab_g2
Sensorex, Operating Dissolved Oxygen Probes in Bioreactors and Cell Culture -
https://sensorex.com/operating-dissolved-oxygen-probes-bioreactors-cell-culture/ [VN] -
read
“It’s possible for air bubbles to collect on the tip of the sensor or to cross over it.” Confirms
the failure mode the DO port’s lean is for. Gives no angle for it.
-> scad/head.scad do_probe_port_tilt_degrees
No retrievable source gives a mounting ANGLE for a DO probe. Mettler-Toledo’s bioreactor DO
sensor page and Hamilton’s pH installation note both surface in search saying vertical installations
are the troublesome case, and neither could be fetched to confirm it - HTTP 406 and an empty
response respectively. So the 4.5 deg the DO port leans is reasoned, not cited. It is also moot:
the jar’s mouth caps it at 4.5 deg and head() asserts that, so the geometry decides the number and
the sources only decide its direction.
Purchased parts
Section titled “Purchased parts”Vendor pages and datasheets are cited inline on the registry row that uses them, since the row is the thing being justified:
- bayonet-port-scad’s
purchased/atlas_probes.scad- Atlas Scientific product pages and datasheets, one per row scad/purchased/vessels.scad- the jar listings each vessel row was measured fromscad/purchased/strip_lights.scad- LED strip listings- Procurement and
purchased-parts.csv- McMaster part numbers with their specifications
Notable rows: bearing 6153K71 (trade no. 608-2RS, and NopSCADlib’s BB608 carries the same
22 x 7 x 8), heat-set insert 97163A152, rim gasket sheet 8525T65 (1/16 in = 1.5875 mm, 60A,
which sets the ASME gasket factor), port o-ring 8785N383.
Software and libraries
Section titled “Software and libraries”- OpenSCAD 2026.10.15, the snapshot CI runs (
.github/workflows/check.yml); 2021.01 is the oldest supported - NopSCADlib - vitamins: screws, nuts, inserts, ball bearings, shaft couplings, extrusions
- bayonet-lock-scad - the bayonet coupling this project’s ports are built on
scad/custom/impeller.scadis adapted from https://infinityplays.com/3d-part-design-with-openscad-57-a-universal-propeller-impeller-design-module/ - a geometry module, not an engineering source
