A batch is finished when four independent signals agree, not when the clock runs out. Here is what to watch, and what each signal can and cannot prove.
The end point is a decision, not a reading
A vacuum dryer has no single gauge that says the batch is done. The end point is a judgement built from four signals that fail in different ways: bed temperature, vapour flow, agitator load and a drawn sample. When they agree, the batch is finished. When they disagree, one of them is misleading you, and finding out which one is cheaper than shipping wet product.
The reason is physical. Under vacuum, water leaves at a temperature set by the absolute pressure rather than by the jacket. As free moisture disappears, the remaining water sits inside the particle and must diffuse to the surface before it can evaporate. That transition changes every signal at once, which is why one number is unreliable and a set of them is not.
Fix the target moisture and its basis first
Nothing can be measured until dry is written as a number with a basis. Half a percent of moisture means two different things on a wet basis and a dry basis, and a specification that omits the basis will be argued about at acceptance. Fix the target, the basis, the method and the sample size before the first batch, and put all four on the batch record.
The method matters as much as the number. Loss on drying in an oven drives off everything volatile, which is correct for a water-wet inorganic filter cake and misleading for a solvent-wet resin or for a hydrate that releases water above a certain temperature. Karl Fischer titration measures water specifically. Choose the method that matches the reason a customer would reject the product.
Target moisture: what the specification has to state
| Item | Why it has to be written down | Typical failure if it is not |
|---|---|---|
| Basis (wet or dry) | The same water content gives two different percentages, and the two are not interchangeable | Batch passes in-house and is rejected on receipt |
| Method (oven LOD, Karl Fischer, gravimetric) | Volatiles other than water respond differently to each method | A solvent-wet batch reads dry while still holding solvent |
| Sample size and sampling point | A two-gram grab sample and a bulk composite describe different populations | A good sample from a good spot hides a wet core |
| Acceptance rule (single value or mean of n) | One result and a mean with a spread are different contracts | Dispute over a single out-of-spec result |
Source: Ruibo dryer commissioning practice. Method selection follows the product release specification, not the dryer supplier.
Signal one: bed temperature closing on the jacket
While free moisture is evaporating, the bed stays near the boiling point of the liquid at the working vacuum and draws heat away as fast as the jacket supplies it. Measure the bed, not the jacket outlet. Once the free moisture is gone, the same heat input has nowhere to go and the bed temperature climbs toward the jacket temperature. The approach is the signal; the absolute value is not.
On a ZPG the vessel pressure is specified from -0.096 to 0.15 MPa, so in normal drying duty the absolute pressure is a few kilopascals and water boils well below 40 degrees Celsius. A bed temperature that has been flat for hours and then starts to rise is reporting that the evaporation load has collapsed. It is the earliest useful indicator and the easiest to misread, because a jacket change or a pressure change moves it too.
Signal two: vapour flow and the vacuum system
Vapour leaving the vessel has to be condensed or pumped, so its flow is a direct measure of evaporation rate. On a condenser-equipped system the condensate rate is the most honest number on the plant: it can be weighed or read from a receiver level, and sensor drift cannot fake it. When condensate flow falls to a small fraction of its early-batch value while the jacket stays hot, the batch has little left to give.
Vacuum level alone is a weak signal and is often misused. A stable vacuum only says the pump is keeping up with the leak rate plus the vapour load. If the vacuum improves sharply with the jacket unchanged, the vapour load has dropped, which is useful. If it degrades, look for a shaft seal, a blinded filter or a condenser that is no longer cold enough before blaming the material.
The vapour outlet on a ZPG carries configurable filter media, which exist to stop fine powder reaching the pump. A blinded filter looks exactly like a wet batch: vacuum falls, evaporation slows, and the operator adds heat. Check the differential pressure across that filter before changing the recipe.
Signal three: agitator load
The rake exists to keep fresh material against the hot wall, and the torque it needs depends on how the bed flows. A wet paste smears and climbs; a dry powder crumbles and slides. On most installations the motor current traces a curve across the batch, and the shape of that curve at the end is repeatable for a given material.
Treat it as a confirmation, not a primary measurement. Current also moves with fill level, jacket temperature and blade wear, so the same absolute value means different things in different batches. What transfers between batches is the change in slope relative to that batch's own earlier plateau.
Stirring speed across the nine standard ZPG models is 8 to 18 rpm, and torque at those speeds is dominated by the material's resistance rather than by inertia. That is why the load signal is usable at all on a machine this slow.
Signal four: the sample, drawn without letting the batch down
A sample is only representative if it comes from a moving bed and stays sealed. Drawing from the discharge port of a batch that has been sitting gives you the driest material in the vessel. Draw from a sampler in the circulating bed, seal the container immediately, and let it cool before weighing so that moisture does not flash off in the balance room.
Sample where the rake actually moves material. On a horizontal ZPG the rake forms a circulation loop and clears the hot cylinder wall, so material near the shaft and material at the end plates sit at different points on the curve. Two sealed samples from two positions answer a question that one sample cannot.
Reading the four signals together
The value of four signals is that they fail differently. Temperature can be fooled by a jacket change, condensate by a cooling-water leak, agitator load by fill level, and a sample by where it was taken. A table of disagreements is therefore more useful than a table of agreements.
When the signals disagree
| What you observe | Most likely cause | What it rules out |
|---|---|---|
| Bed temperature rising, condensate still flowing | Bed is drying unevenly: the wall zone has dried while the core has not | A genuinely finished batch |
| Condensate stopped, bed temperature flat | Vapour path restricted by a blinded outlet filter or an undercooled condenser | End of drying |
| Vacuum improves sharply, bed temperature climbs fast | Free moisture gone; sensible heat now raises the bed | A need for more jacket heat |
| Agitator load drops, sample still wet | Sample came from a dried surface layer, or the bed has segregated | Homogeneity of the batch |
| Sample dry, load unchanged | Sample too small, or taken from the wall film rather than the bed | That the bulk is at target |
Source: Ruibo operating practice for agitated vacuum dryers. Confirm each row against your own batch records before using it as a rule.
Why the last few percent take the longest
The tail of a vacuum batch is limited by diffusion inside the particle, not by heat supply. Adding jacket temperature at this stage raises the product temperature and the risk of thermal damage without shortening the tail meaningfully. The levers that still work are time and residence, plus a lower bed depth so the vapour has less distance to travel.
This is the stage where an operator is most tempted to declare the batch finished, and where the four-signal rule earns its keep. If the bed temperature has climbed but condensate flow has not fallen, the batch is not finished. It is being overheated.
What to put on the batch record
A record that lets the next batch be shorter is worth more than one that only proves compliance. Log jacket temperature, bed temperature, absolute pressure, condensate mass and agitator current at fixed intervals, and record each sample with its position and method. After three batches the shape of those curves is a recipe; after one it is only a story.
- Jacket inlet and outlet temperature: shows the real driving force, not the set point.
- Bed temperature at one fixed point: the trend carries more information than the value.
- Absolute pressure: gauge vacuum hides the difference between 5 kPa and 20 kPa absolute.
- Condensate mass per interval: the closest thing to a direct evaporation-rate measurement.
- Agitator current: the confirmation signal, read as a change in slope rather than a level.
- Sample result with position and method: without both, the number is not reproducible.
Questions engineers ask about the end point
The same five questions come back from every commissioning team. The answers below apply to agitated vacuum dryers in general and to the ZPG range in particular.
- Can the vacuum gauge alone tell me the batch is finished?
- No. A steady vacuum only shows that the pump is handling the leak rate plus the vapour load. The gauge becomes useful when the vapour load drops and the vacuum improves sharply at an unchanged jacket temperature, and even then it needs the bed temperature and the sample to confirm.
- How often should a batch be sampled?
- Sample on a fixed interval that shortens as the batch approaches target, and always sample from the same two positions so the results are comparable between batches. Sampling only at the end produces a number without a trend, and a number without a trend cannot tell you whether the next batch will finish sooner or later.
- Does a long tail mean the dryer is undersized?
- Not necessarily. A long tail is usually diffusion-limited drying, which no amount of jacket area removes. If the tail responds to a lower bed depth or to a longer residence at the same temperature, the machine is doing its job; if it responds only to heat, the heat transfer area is worth re-examining.
- Is Karl Fischer always better than oven loss on drying?
- It is more specific, not universally better. Karl Fischer measures water and is the right method when the product can hold solvent or when a hydrate would decompose in the oven. Oven loss on drying is faster and cheaper and is adequate for a water-wet inorganic material with no volatile component.
- Can the end point be automated on a vacuum rake dryer?
- Partly. Bed temperature, absolute pressure, condensate mass and agitator current can all be trended and given alarm limits, and a combination of them can stop the batch automatically. The sample still has to be taken, because none of the four signals proves the moisture content; they only show that the batch has stopped changing.