How to Evaluate Saggar Deformation During High-Temperature Sintering

2026-10-21

Saggar dimensions can change in service. A saggar that seated cleanly on the setter at the start of a campaign may sit differently after extended firing — and in processes where dimensional consistency matters, that change can eventually affect the product.

Many plants already measure something. Caliper readings at shift change, a height check at unloading, an occasional flatness check on the contact face. The measurements exist. The problem is that they rarely support a decision: nobody can say whether a reading is real, what it should be compared against, or what should happen when it crosses a line.

The value of deformation management does not come from describing the phenomenon. It comes from two things: a measurement that can be repeated, and acceptance criteria that were agreed before the data arrived. This guide is organized around four actions that connect them:

Measurement → Recording → Comparison → Evaluation.

Each action answers a question the previous one raises. Skip one, and the readings stay observations instead of decisions.

Why deformation data often fails to support a decision

When a dimensional change appears in the log and the team still cannot act on it, the cause is usually not the instrument. It is one of three missing definitions.

Three reasons measurement data gets set aside

The measurement position drifts. Nobody defined where on the saggar to measure. One operator measures the outer wall at mid-height, the next measures near the rim, the third measures the contact face. Three different numbers enter the same column, and none of them can be compared with the next round.

The method changes between measurements. A reading taken hot on Monday and a reading taken cold on Thursday are different numbers, even on an unchanged saggar. Changing instruments, operators, or the waiting time after unloading has the same effect: the log ends up recording the method as much as the saggar.

No acceptance threshold was agreed in advance. The number arrives, and the discussion starts. Is it a lot? Compared to what? Does it matter? The debate takes longer than the measurement, and by the time a position forms, the saggar has fired several more cycles.

Without these three, a dimensional reading stays a trend observation. The rest of this guide builds the four actions that close the gap.

Step 1 — Make the measurement repeatable

Repeatable means the same saggar, measured by different people on different days, gives the same answer within a known margin. Repeatability is a property of the procedure, not of the instrument.

Defining measurement positions and reference method

Choose measurement points that relate to function, and mark them on the saggar so they survive use:

  • The contact face that carries the workpiece — usually the surface where flatness matters most.
  • A fixed height on the outer wall, referenced to the bottom or the contact-face plane.
  • Defined corners, or diametrically opposite points, for height and diameter checks.

Then define the datum: measure from the contact-face plane, or from the bottom face — one reference, written down, used every time. Measuring from whatever surface is convenient reintroduces the drift this step exists to remove.

Choosing the measuring instrument and its resolution

The instrument must resolve the change you intend to detect. If the boundary you care about is a fraction of a millimetre, an instrument read to the nearest millimetre cannot support a judgement against that boundary.

Two practical rules: keep the same instrument type for the life of the comparison, and record the instrument ID in the log. When an instrument is replaced, note the change and check the new one against the old on a reference piece before the series continues.

Fixing the measurement condition: temperature and timing

Dimensions move with temperature. A hot reading and a cold reading on the same unchanged saggar are different numbers, so the measurement condition is part of the method, not a detail.

Fix it explicitly:

  • Measure at room temperature after full cooling, or at a defined point in the firing cycle — one condition, stated in the procedure.
  • Fix the waiting time between unloading and measurement.
  • If hot measurement is unavoidable, keep those readings as a separate series. Never mix hot and cold readings in one comparison.

Step 2 — Define what to record

Recording is where deformation programs often quietly fail: the measurement happens, but the record cannot answer questions later.

Dimensional change, weight change, warpage, contact-face flatness

Define the record items before the first measurement:

  • Dimensional change at the agreed points — the primary record, taken against the datum from Step 1.
  • Warpage and contact-face flatness — in many sintering applications these matter more than overall dimensions, because they determine how the workpiece sits and how heat transfers at the contact face.
  • Weight change — useful as an auxiliary observation. Weight loss or gain can point to oxidation, deposits, or material loss at the surface. It is not a geometric deformation indicator, and it must not be used as a substitute for dimensional measurement: a saggar can lose weight without changing shape, and it can change shape without losing weight.

Recording convention and units

A record that survives beyond the shift needs a fixed format:

  • One log per saggar, one row per measurement event.
  • Every row carries: date, cycle count or campaign identifier, operator, instrument ID, measurement condition (hot or cold, waiting time), and the raw readings.
  • Record the raw reading, not only the calculated change. Deltas can be recalculated later; raw readings cannot be recovered.
  • Fix the units and the number of decimal places at the start, and do not change them mid-series.

Step 3 — Set the comparison basis

A number means nothing until it has something to be compared against. The basis has to be chosen deliberately, because different bases answer different questions.

Against the first-cycle reference or against a specification

Two bases cover most needs:

  • First-cycle reference — compare each reading against the same saggar's first-use or as-received measurement. This basis shows cumulative change in that individual saggar.
  • Specification — compare the reading against the drawing or agreed dimensional tolerance. This basis shows conformance at the moment of measurement.

Pick one primary basis for each decision, and state it in the log. Mixing the two in one column makes the series unreadable.

What a comparison can and cannot tell you

A first-cycle comparison tells you the direction and size of the change. It cannot tell you whether the change is acceptable — that comes from the threshold, which is Step 4.

A specification comparison tells you pass or fail at this moment. It cannot tell you the rate of change, and rate matters for planning: a saggar drifting toward the limit at a steady rate can be scheduled for replacement before it becomes a problem. The same saggar found just inside the limit after a sudden step change is a different situation, and the comparison alone will not show it.

Step 4 — Set the acceptance criteria in advance

This is the step that converts data into decisions, and it is the step too often left until after the first surprising reading.

Defining rejection thresholds before production starts

The threshold belongs to your process. Derive it from what your line can absorb: how much saggar dimensional change affects the workpiece, the stacking arrangement, the handling, and the furnace geometry before quality or yield is at risk.

There is no universal deformation figure, and any article or supplier offering one is offering a convention, not a rule. The right threshold depends on your part tolerances, your loading arrangement, and your furnace. This guide deliberately does not propose a number.

What matters is that the threshold — and the measurement that triggers it — is written into the saggar management procedure before the campaign starts.

Responding to a threshold breach

What happens when a reading crosses the threshold is a process decision, not a measurement decision. The options usually include investigating before acting, increasing the monitoring frequency, removing the saggar from service, or stopping the affected production.

Which option applies depends on your own process requirements and risk assessment. A line running a tight-tolerance product may pull the saggar immediately; a line with margin in the process may investigate and monitor first. No single response is universally correct, and stopping production is not automatically the right answer. What matters is that the response was decided in advance — so the reaction follows a procedure instead of starting a debate.

Step 5 — Separate real deformation from other effects

Not every dimensional reading is deformation. Two effects routinely masquerade as it, and one real effect is routinely dismissed.

Distinguishing from thermal expansion and from measurement error

Thermal expansion is reversible and follows the measurement temperature. If readings are taken hot, part of the apparent change is simply temperature — control the measurement condition from Step 1 rather than correcting for it afterwards.

Measurement error comes from the instrument, the reading, or the surface: resolution limits, parallax, the wrong mark, a contaminated contact face. A single surprising reading is a reason to re-measure, not a reason to act.

Warpage caused by loading is still real deformation

Warpage caused by how the saggar is loaded — stacking weight, support arrangement, contact with adjacent kiln furniture — is not an artifact. The saggar has genuinely changed shape, and that shape change reaches the workpiece the same way any other deformation does.

So do not exclude loading-related warpage from the record as "not real deformation." Record it as a dimensional change, and treat the loading arrangement as the first line of investigation: stacking height, load distribution, and support condition.

The distinction that matters here is cause, not real versus not real. If the change traces to the loading arrangement, replacing the saggar alone will not resolve it — the loading has to change as well.

Summary

Deformation management comes down to four actions, in order, none of them optional:

  1. Measurement — fixed positions, a stated datum, a defined instrument, a fixed measurement condition.
  2. Recording — defined items, fixed conventions and units, raw readings kept; weight change as an auxiliary observation only.
  3. Comparison — a stated basis, with its limits understood.
  4. Evaluation — thresholds agreed before production, and responses decided in advance.

A dimensional reading is a measurement result, not a verdict. Whether a change matters is decided by the acceptance criteria your process agreed before the data arrived — and by whether the four actions above were in place to turn readings into decisions.