Behavioral techniques: from quality control to instrumented micro-pilots

The behavioral instrumental techniques (Cup of viscosity, viscometer, texturometer, rheometer, tribometer, etc.), occupy a privileged place in our Industrialology, in that they make it possible to acquire information relating to the macroscopic scale, characteristic of industrial phenomena.

Among the wide variety of devices on the market, it is often difficult for the future user to identify the potential of these techniques. To this end, I would like to analyze their modes of operation, which I hope will allow us to better understand the potential and limits of their use.

“Measurement-Sensor” mode

The instruments of an analysis laboratory are as many more or less voluminous, more or less sophisticated sensors. Behind different scientific principles of measurement, current instrumentation works on a single principle: it assigns a value to the sample or the measurement environment (a temperature, a pressure, a chemical content, etc.).

We find the definition in theInternational Vocabulary of Metrology: the measure (the “Measurement”) is the “process consisting of experimentally obtaining one or more values ​​that can reasonably be attributed to a quantity”. Already in 1920, Norman Campbell, pioneer of measurement theory, defined it as follows: “Measurement is the assignment of numerals to represent properties”.

This mode of operation, I was led to give it the name of "Measurement-Sensor" to specify that it corresponds to a particular mode of operation of the measurement, not universal, and thus to differentiate it from other modes on which I will come back later.

Set measurement conditions to produce a single value

If we look at the internal conditions of the instrumentation to produce a single value, we see that they require a determined and fixed measurement process. This is moreover precisely the purpose of the prescriptions of the standardized tests, which generally fix the type of device, its possible adjustments and the implementation protocols, which can include the preparation of the samples.

In practice, the measurement process is sometimes imposed by the very physical structure of the instrumentation. In the case of the most rudimentary behavioral techniques (viscosity cup for fluids or Hall funnel for powders), the sample flows by gravity through an orifice in a device of fixed geometry and material and a single quantity is determined (flow time for the viscosity cut, slope angle or flow index for the funnel).

Powder Slope Angle Tester (from Coppley Scientific)

In slightly more complex devices, such as the simple viscometer -of the Brookfield type-, the instrumentation is designed with a few degrees of freedom additional in the implementation of the material (container size, choice of mobile, speed of rotation), but its use in Quality Control most often consists of fixing them to determine a viscosity value.

The representativeness of the single point

In the case of viscosity, I have already had the opportunity to mention at length in the first articles that the rheological behavior of a product is generally not reduced to a viscosity value: viscosity is a function of the conditions of product implementation.

The practical consequences are considerable: a single viscosity value is generally not representative (except for Newtonian products) of the rheological behavior of the material. But what is it to be representative? This means that comparing the measured values ​​for two samples is enough to transpose the comparison to differences in actual behavior.

Pen ink flow curves (note in particular that these curves intersect around 10 s-1, corresponding to the Brookfield viscosity test range)

One could easily justify the same criticism at the angle of the slope of the powders and many other normative parameters, whose capacities to discriminate between materials and products whose behaviors are nevertheless drastically different, are faulted in all industrial sectors.

The limits of the simple view of measurement

To sum up, thea common sense view of measurement (like Measure-Sensor, i.e. on the model of the thermometer) is not sufficient to account for the complexity of material behavior experienced on an industrial scale. Worse: it misleads men and women who believe in good faith (why believe anything else?) that the values ​​of technical sheets or quality control are duly representative of what makes the difference between materials.

It took us several years to document this state of affairs applicable to all the industrial sectors that we had the opportunity to meet. And for good reason: if the static characteristics of matter were sufficient to predict its behavior, then modifying the conditions of their transformation would have no impact.

This observation, at first sight discouraging, is on the contrary rich in promises of possible improvements, provided we take a little distance vis-à-vis common sense and pseudo-evidence to seek to understand what is and can be used for measurement.

The context of the measure, a detail?

According to the usual meaning of measurement, it is above all an operation: assigning values. What to measure, how, with what device, how to interpret results, would not strictly speaking concern the operation of measurement, but a gray area in the contours of the project that motivates said measurement.

However, in the daily life of the operator, these questions are crucial, as critical as the measurement operation itself. The best proof, if any, lies in the observation in the field of the difficulties encountered when the measurement is not consistent with the empirical observations. How and why the values ​​measured for two products, one of which poses problems in the process, the other not, can be equivalent. What justifies carrying out the measurement in such and such a way?

In difficulty, measurement becomes part of the problem, but in a very different form from questions of precision or uncertainty which are often seen as the only relevant issues. Taking a little height on the context of any measure will allow us to better understand what it achieves in practice.

To measure is to ask questions

We sometimes tend to forget it, but a measure is always carried out in a concrete context with a given objective - whether it is a question of checking the conformity of a parameter with specifications, of specifying environmental conditions (what temperature is it in the workshop?), of determining differences in product characteristics, of comparing inputs , to understand phenomena, etc.

To measure is always to carry out an experiment with instrumentation, that is to say to ask questions of matter. Once again, the restrictive vision of the Measurement-Sensor and the imago of the analysis laboratory which measures the properties of samples without knowing anything about them or the associated problems is a source of confusion.

Measurement is always guided by a concrete problem, for which the quantification of certain properties (to remain within the framework of the classic definition of measurement) serves to establish classifications, interpretation grids, models of the concrete situation. The quality engineer, the formulator, the designer, rely on measurements for various needs.

In other words, measurement systematically serves an intellectual activity linked to concrete problems and objectives. These intellectual activities in the abstract universe of numbers aim to feed concrete actions: orient them, compare them, validate them, suggest them, etc. Thus, a fortiori in the industrial world, abstraction is only a passage, not an end in itself.

This is where another mode of using instrumentation than that of Measurement-Sensor takes on its meaning: Measurement-Experience.

The “Measurement-Experiment” mode

I summarized the current mode of operation in "Measure-Sensor" on the basis of a simple principle: one sample = one value.

In the “Measurement-Experiment” mode, a sample will be able to give rise to a whole series of values. This is subject to modifiable and controlled measurement conditions, giving rise to the production no longer of a single value but of a set of values ​​according to the conditions imposed. We are talking about functions, or response curves.

In the case of behavioral techniques, the modifiable measurement conditions are generally thermal and mechanical constraints (shear, compression, etc.) or environmental constraints. Rudimentary techniques generally do not allow such modifications to be made, making the Measure-Experiment mode inaccessible.

Among the standard instruments of viscometry and rheometry, there is a wide variety of degrees of freedom in the measurement conditions, ranging from discrete adjustment of spindle rotation speed for the simplest to extensive control of stress, shear, thermal conditions, etc. Obviously, the cost of instrumentation generally increases with the parametrability measurement conditions, both in their number and their amplitude or precision.

Understand the impact of influencing factors

Where the Measure-Sensor mode limits the interpretable influence factors to the conditions of preparation of the samples – for which it is a question of being able to guarantee a fine enough control to hope to interpret them –, the "Measurement-Experiment" mode opens up the possibility of studying the influence of controlled parameters on the same sample.

This is how the production of "flow curves" in rheometry -and certain viscometers- gives an understanding of the behavior profile of matter, which can be interpreted in different real contexts. For fluid products, the low shear zones will be rather representative of behavior at rest or of limited movement, whereas the high shear zones will be characteristic of stronger flow. For powders, it becomes possible to determine the tendency to flow under variable compression conditions - information inaccessible by a simple slope angle measurement.

It becomes possible to study the influence of variations in these stresses on the material, for example the impact of thermal or mechanical conditions. Obviously, these variations come on top of the variations likely to be imposed on the samples themselves. The study approaches are therefore not reduced to experimental plans for measuring fixed parameters on a given sample, but to an experimental approach in which the access routes to information on the behavior of matter do not are not fixed once and for all.

From qualitative trend to quantitative prediction

Through this Measurement-Experiment mode, the interpretation of the behavior of the samples is enriched in a considerable way, allowing a behavioral understanding as fine as the controlled variabilities allow.

The interpretations can remain of a "qualitative" order, that is to say translate trends of variations to increases, decreases in the control parameters. These trends are often enough to determine the favorable or unfavorable conditions of implementation (the impact of such thermal conditions in a process byextrusion for example).

The measurements can also be used more quantitatively in the context of modeling the product/process coupling, for example in the context of determining the flow rates/pressures of fluid products in the pipeline.

The variety of experimental control conditions also allows the implementation of predictive protocols, by adjusting the measurement conditions to the stress conditions experienced by the material.

The implementation of predictive protocols

Unlike the Measure-Sensor mode, the Measurement-Experiment mode allows you to "play" on the measured parameters, the protocols implemented to produce them -but also the post-processing of the parameters- to adjust the representativeness of the results obtained in relation to the real behaviors observed. The objective is to determine descriptors at the same time as the way to quantify them.

A descriptor can be a parameter or more or less complex combinations of parameters whose characteristic is to satisfactorily represent the real variations observed in a concrete situation.

Material behavior in industrial processes or applications can thus sometimes be predicted by simple protocols and parameters, where others require the implementation of several tests and in some cases post-processing of the parameters measured.

Measurement-Experience thus offers R&D this predictive, even prescriptive capacity. Once the descriptor(s) have been established, it is possible to set up the corresponding Measurement-Sensor approaches -on less configurable equipment- to simplify routine measurements.

The "Instrumented micro-pilot" mode, extension of the field of experience

It is possible to go even further in the mimicry of the real conditions of implementation, through the approaches of the type " Instrumented Micropilot ».

Sufficiently sophisticated behavioral instrumentations provide a robust and accurate measurement basis for many properties and stress conditions. In addition, certain adaptations make it possible to make real micro-pilots, in which it is possible to come and incorporate various ingredients, air, to test geometries of mixed.

It is thus possible to transform fluidized bed powder measuring cells into micro-granulators, micro-emulsifiers, adapt injection test benches, etc. The addition of external sensors makes it possible to combine optical or other measurements with thermomechanical and rheological measurements, further enriching the spectrum of relevant descriptors to represent real situations.

Explore experimentally the couplings between material and conditions of implementation

These different measurement approaches based on behavioral techniques thus offer the possibility of studying and understanding the couplings between material, process and application while minimizing the consumption of material, equipment and human resources..

It thus becomes possible toevaluate optimizations of an existing process without mobilizing the production tool (or in a limited way), sizing the equipment adapted according to the variety of products,optimize formulations with regard to process or application constraints,evaluate candidate ingredients or visit avenues of innovation.

The history of industrial development has led the R&D departments to act according to the logic of quality control. At a time when it is a question of being able to innovate quickly, many realize that all these controls do not provide any understanding allowing to direct developments in the right direction.

Behavioral techniques provide many answers to those who take the trouble to exploit their potential and there is no doubt that the effort to be made is negligible compared to the direct and indirect costs induced by trial/error approaches, just as the return on investment is considerable.