A Reproducible Protocol for Researchers and Engineers

Ask a question the experiment can answer

The protocol turns a design uncertainty into an interpretable comparison. Its purpose is to test the proprietary commercial Aguas SB T1 method, not to present the activity as a university programme or turn the experiment into a promotional demonstration. A campaign should begin with one primary question: transfer across scales, exposure response, separation effects or predictive capability of a model version. Accumulating objectives without prioritization often produces many measurements and few identifiable conclusions.

The question becomes a primary outcome, comparison and acceptance criterion before responses are observed. For example, investigating intermediate separation requires configurations with equivalent exposure and records of removed material. Investigating prediction requires inputs, version and rules to be frozen before held-out results are opened. Criteria should express engineering relevance and required precision, not merely statistical significance selected afterwards.

Operational preregistration

The prior record identifies the responsible person, date, T1 2.3 version, effective configuration, units, question, hypotheses, and primary and secondary variables. It describes batches, treatment allocation, replicates, sampling points, permitted exclusions and handling of censored values. It includes the analysis plan and conditions that would make the test inconclusive. It may remain under commercial document control; its purpose is establishing precedence and traceability, not disclosing confidential information.

Three forms of acceptance should be separated. Operational acceptance confirms that the setup functioned within permitted limits. Analytical acceptance confirms that laboratory controls met agreed criteria. Hypothesis acceptance uses valid comparison results. Failure of the first or second may prevent evaluation of the third; it does not demonstrate that treatment works or fails. Deviations are recorded and analysed according to their implications.

Inlet water, batch and preparation

Each batch needs an identifier, an origin description without publicly identifying details, and collection, storage and preparation conditions. Initial characterization follows the question: total and soluble COD, TOC, solids, total N and nitrogen species, P, pH, conductivity, alkalinity and temperature where relevant. Measuring everything in every test is unnecessary; justifying that omitted variables do not prevent interpretation of the primary outcome is essential.

Homogenization should produce comparable aliquots without unnecessarily changing the matrix. Prior settling, sieving, filtration, dilution and additions are recorded. Differently prepared samples constitute different conditions even when they originate in the same container. Synthetic water helps isolate factors; real water supports transferability investigations. Neither universally replaces the other. The selected composition and its limitations are part of the evidence.

Hydraulics and exposure before treatment

The circuit is drawn with tank, feed, recirculation, purges, modules and separators. Working volume and actual flow are measured under operating conditions. Parallel branches and flow distribution are identified; in series, common flow is retained and module contacts counted separately. Batch exposure is c=Qr*t/Vb under constant conditions; changes in flow, volume or active periods require integration and time records.

Where sample correspondence depends on mixing and residence, a tracer experiment compatible with the installation and matrix is proposed following a safety assessment. RTD informs AF/EF sampling windows; it is not a retrospective correction for every discrepancy. Nominal time V/Q does not itself demonstrate the absence of short-circuiting. Hydraulic and chemical tests must be distinguished if the tracer changes relevant conditions.

Example 1: A batch exposure series

The first proposed design uses a homogenized batch and points c=0,1,2,4,8, adding others where the question requires. This series is neither universally optimal nor a dose recommendation. In a synthetic setup with Vb=0.75 m3 and Qr=3 m3/h, corresponding times are 0, 0.25, 0.5, 1 and 2 h. These are scheduling calculations, not efficacy results.

A baseline is taken before treatment and each sample volume is recorded. If withdrawals materially change Vb, integrated exposure is updated or independent reactors are used for each point. Successive observations from one batch form a correlated series, not five independent replicates. If the programmed twelve-cycle jump is investigated, points near that boundary are added without assuming a physical threshold.

Energy, electrical charge and material

The time record includes power or integrated energy, current and voltage in the electrochemical circuit, electrode condition, temperature, pressures, gas flow and reagents where present. An explicit auxiliary boundary is retained. Specific energy uses net treated volume; electrical charge uses integrated current and coherent units. Faraday's law and its material interpretation are developed in Chapter 04 without turning theoretical mass into measurement.

Residual liquid, samples, sludge, foam, replenishment and separations are accounted for. Before the experiment, the protocol defines which streams can be analysed and which balance closure will be possible. Useful aqueous removal does not require proof of mineralization, but it must be named correctly. A mineralization or nitrogen-transformation hypothesis requires specific fate observables, not only percentages disappearing from water.

Example 2: Hydraulic control and dose-matched separation

The second design divides comparable batches between recirculation without intermediate separation and two treatment periods with solids removal between them. As far as the setup permits, total exposure, electrical charge, active dose and thermal conditions are matched. Any energy or hydraulic difference introduced by separation is measured. An additional hydraulic control preserves circulation without activating the investigated operation, whenever technically feasible and safe.

If matching every dose is impossible, the actual comparison is declared: complete configurations, not a pure separation effect. Whole-water and soluble analyses, separated material and resources are examined. Branch differences may support an operational hypothesis but do not establish synergy across every mechanism. An interaction requires a design also capable of estimating the relevant contributions under comparable conditions.

Replicates, order and practical statistics

An independent replicate repeats the relevant experimental unit; a second reading of the same aliquot evaluates another source of variation. Batch, reactor, run, sample and reading must be labelled separately. A pilot with, for example, three independent runs per condition can inform variability and feasibility but does not guarantee statistical power or sufficient precision. The eventual number depends on the minimum relevant effect, variability and dependence structure.

Order is randomized where possible and blocked by batch or day where those differences matter. This avoids confounding treatment with progressive heating, wear or sample ageing. Analysis reports effect magnitude, dispersion, sample size and relevant individual outcomes. An isolated p-value does not replace technical relevance or matrix coverage. Losses and failures remain in the tracking record.

Example 3: Held-out prediction and acceptance

The third design separates complete batches for development and evaluation. Before the held-out group is revealed, inputs, outputs predicted by T1 2.3 and acceptable-error criteria for the engineering decision are archived. Concentration, load where relevant and energy are compared while preserving analytical limits. No universal tolerance is set here: it must reflect the measurement method and intended use.

If prediction fails, data quality, hydraulic correspondence and model domain are examined. A subsequent explanation may generate a new hypothesis but does not turn the same held-out set into independent validation of a refit. Any future calibration is versioned and requires another evaluation. An inconclusive test is reported as such, without selectively retaining favourable analytes or conditions.

A bounded first comparison and decision criterion

The first comparison can combine the preceding designs without attempting to validate every mechanism at once. Select a matrix, configuration and primary response; record a development batch and reserve another independent batch to evaluate the prediction before refitting. Compare circulation control, activated train and matched separation only where the equipment can safely distinguish them. Do not construct nominal controls that change every condition simultaneously.

Define the decision in advance: reproducing the curve supports that representation only under tested conditions; material bias requires restricting the domain or proposing another parameterisation; insufficient hydraulic, analytical or energy data make the question inconclusive. A concentration pair may show removal but cannot simultaneously identify rate, ceiling and separation effects. Do not fit every factor to that same pair.

This comparison does not validate all matrices, sizes, chemical options or serial configurations by extension. Its value is resolving a priority uncertainty and providing a transferable result with limits, not turning a favourable pilot into universal validation.

Safety, quality and reproducible delivery

Operation requires competent assessment of electrical, pressure, gas, chemical and waste hazards. The protocol does not prescribe improvised reactive mixtures or atmospheres. The laboratory defines preservation, oxidant neutralization and interference handling according to its methods; blanks, positive controls and duplicates check analytical quality. Vapour cavities and air bubbles are characterized differently, and pH observations do not automatically change the engine.

The delivery includes protocol and deviations, schematic, original and normalized data, unit dictionary, configuration, energy and material records, exclusion rules and comparison results. Public reporting separates synthetic scenarios, authorized measurements and model outputs. The reproducibility bank verifies frozen execution; the experiment tests hypotheses under specific conditions. The valuable outcome is a repeatable conclusion, even when it refutes a commercial expectation or defines a smaller domain than anticipated.