Crystallizer jacket cooling/heating loop: keeping the jacket close enough to the batch that a cooling ramp does not crystallise product on the wall

01

From 60 °C to 20 °C on a set trajectory: the nucleation approach, linear and cubic ramps the loop has to track

Chew and co-workers (Institute of Chemical and Engineering Sciences, Singapore, with AstraZeneca) ran glycine cooling crystallizations in a 500 ml jacketed flask stirred at 550 rpm, with a Julabo HP50 circulator pumping deionised water through the jacket and a Pt100 reading the contents every 2 seconds. [1]

For glycine the authors found hardly any difference in the final chord-length distributions between natural cooling, cubic and linear profiles; they attribute this to glycine's fast growth, estimated at 62 nm/s at 0.3 °C/min, which makes nucleation the controlling step. [1] The same paper notes that crystallizations above 5 litres are conventionally run on a pre-set cooling profile taken from small-scale experiments, and states the paper's purpose as testing whether newer control methods give the more consistent product and easier scale-up claimed for them. [1]

Editorial calculation from the set points above: a linear ramp from 60 °C to 20 °C takes 40 K ÷ 0.7 K/min ≈ 57 minutes at 0.7 °C/min and 40 K ÷ 0.3 K/min ≈ 133 minutes at 0.3 °C/min, so the loop has to hold a stable ramp for one to two hours, not just reach an end temperature. [1]

State the dissolution, seeding and final temperatures, the ramp rates or profile between them, and whether the controller follows the contents (process) or the jacket outlet.

02

How far ahead of the contents the jacket runs: 6.5 K/min against 2.3 K/min at 20 litres, and wall nucleation after scale-up

Huber's case study CS 92 shows the gap between jacket and contents directly. A Unistat 910w was connected to a 20-litre Buchi Glas Uster jacketed glass reactor holding 15 litres of a silicone-based heat-transfer fluid, stirred at 70 rpm, through two 1.5 m insulated metal hoses, under process control. [3]

Editorial calculation from CS 92: at the start of the ramp the jacket fell about 6.5 ÷ 2.3 ≈ 2.8 times faster than the contents, and at the lowest settled point it stayed 7 K colder (−82 °C against −75 °C). [3] In a crystallizer that lead is the wall running colder than the bulk, which is where unwanted nucleation starts.

Lee, Kim and Bang (Sogang University) describe exactly that at scale-up: because the heat-transfer area per volume falls in a larger crystallizer, the temperature difference between the reactor and the jacket grows and may induce unwanted nucleation on the crystallizer wall. They designed a cooling-holding profile that minimises undershoot of the jacket temperature and tested it scaling from a 100 mL to a 5 L crystallizer. [2]

Editorial calculation, assuming geometrically similar vessels: going from 100 mL to 5 L multiplies the volume by 50 and the linear size by 50^(1/3) ≈ 3.7, so wetted jacket area per litre falls to about 1/3.7 ≈ 27 % of the small vessel's value. [2] To remove heat at the same rate per litre, the jacket must run roughly 3.7 times further below the contents unless the heat-transfer coefficient rises.

Ask for the jacket-to-contents temperature difference the loop will need at your batch size and ramp, and for a limit on jacket undershoot during seeding and the first part of growth.

03

The expansion tank on a single-fluid loop: up to 30 % swell, 25 % cold level, 75 % hot level

Because one fluid both heats the batch to dissolution and cools it to the end point, its volume changes through every cycle. Eastman (Therminol heat transfer fluids) states that fluid volume can expand by up to 30 % on reaching operating temperature, and gives a 25 % cold-level and 75 % maximum hot-level guideline for the thermal expansion tank, for fluid availability and vapour disengagement. [4]

Eastman also places the expansion tank at the highest point of the system for gas collection and pump performance, names it the main venting point for moisture and low-boiling components, and lists level measurement and overpressure relief as the instrumentation that prevents overfilling. [4]

Editorial calculation from the 25 %/75 % levels: the swell has to fit within half of the tank volume, so the tank must hold at least twice the expansion. For an assumed 300 litres of fluid in the jacket, piping and exchanger and an assumed 10 % expansion across the loop's temperature span, the swell is 30 litres and the tank needs at least 60 litres; at Eastman's upper figure of 30 % it would need 180 litres. [4] The fluid volume and expansion figure here are assumptions for illustration; the real figure comes from the chosen fluid's density at the lowest and highest loop temperatures.

Give the total loop volume (jacket, hoses or piping, exchanger, pump), the fluid, and the lowest and highest loop temperatures so the tank, vent and relief can be sized.

04

What to send for a crystallizer jacket cooling/heating loop

Paste it into your enquiry below or into an email to a maker, then replace every bracket.

Vessel: [volume; jacket type; jacket volume; heat-transfer area]
Profile: [dissolution, seeding and final temperatures; ramp rates or profile]
Control: [process (contents) or jacket; allowed jacket undershoot]
Utilities: [chilled water, brine or refrigeration; steam or electric heat]
Fluid: [type; lowest and highest loop temperature; loop volume]
Skid: [pump; exchanger; expansion tank; hazardous-area rating]

See manufacturer records Send an enquiry through CNFX