Two-Phase Gas-Liquid Flow Regimes and Slurry Transport

Scope: Packed beds; fluidization; gas-liquid flow; slurries.

Integrated introduction

Several industrial contactors exchange momentum between phases. Packed-bed resistance involves a viscous contribution linear in superficial speed and an inertial contribution quadratic in speed. A bed begins to fluidize when pressure drop supports the effective weight of solids; beyond that point the settled-bed Ergun equation is not sufficient by itself. Gas–liquid and slurry correlations are regime-dependent and need experimental verification.

Industrial fluid-flow problems require a useful hierarchy. Start with conservation of mass, momentum and energy, then select a constitutive property model, and only afterwards evaluate device performance. A valid equation from an invalid model selection produces an invalid answer. The student should learn to recognize when a phenomenon can be isolated and when the coupling between the fluid, geometry and operating boundary conditions is too strong for a simple correlation.

The unit combines four subjects that should not be learned as unrelated equations. The first establishes the physical quantities, the second shows how they are transported or transformed, the third introduces the influence of an actual process device, and the fourth tests boundaries and uncertainty. Successful engineering interpretation means defending the relative importance of each mechanism, not merely calculating its contribution. Compare results against the appropriate dimensionless parameters and against at least one independent measurement whenever feasible.

Principle 1: Packed-bed friction

Ergun combines a viscous term with a kinetic term. Superficial velocity uses empty vessel cross-section, not the void area. In practical calculations, Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp) provides a compact organizing relationship, but its symbols conceal choices about reference conditions and constitutive data. Before interpreting the output, identify the actual state of the fluid and the geometric scale of the selected section. A flow model should satisfy the limiting condition of zero throughput or zero driving force when that is physically relevant. Such checks can expose a factor-of-four friction convention error or a pressure-reference mistake before any large computational exercise.

Process example. Estimate pressure required to circulate process gas through a catalyst bed. First state the unknown to be solved and identify which measurements can constrain it. Write the full relation and check units before estimating the dominant terms. Particle shape, wall effects and channeling restrict accuracy. If the result is used in a design discussion, accompany it with a sensitivity study rather than an unqualified single-point value.

Counterexample. Inserting the interstitial speed in a superficial-speed correlation distorts both contributions. The lesson is that algebraic correctness is not equivalent to physical correctness. A chosen property, correlation or boundary condition can fail while the numerical method continues to produce smooth and plausible-looking results. Use field measurements to decide whether the assumption holds instead of relying only on the appearance of a simulation graph.

Principle 2: Fluidization onset

Incipient fluidization is approximately the point where fluid drag supports the apparent weight of bed solids. In practical calculations, Δpbed≈(ρs−ρf)(1−ε)Lg provides a compact organizing relationship, but its symbols conceal choices about reference conditions and constitutive data. Before interpreting the output, identify the actual state of the fluid and the geometric scale of the selected section. A flow model should satisfy the limiting condition of zero throughput or zero driving force when that is physically relevant. Such checks can expose a factor-of-four friction convention error or a pressure-reference mistake before any large computational exercise.

Process example. Predict minimum airflow needed to suspend a granular catalyst in a small test column. First state the unknown to be solved and identify which measurements can constrain it. Write the full relation and check units before estimating the dominant terms. Cohesion, irregular particles and bubble formation require more detailed correlations. If the result is used in a design discussion, accompany it with a sensitivity study rather than an unqualified single-point value.

Counterexample. The pressure drop does not keep increasing according to the fixed-bed Ergun equation after a bed expands. The lesson is that algebraic correctness is not equivalent to physical correctness. A chosen property, correlation or boundary condition can fail while the numerical method continues to produce smooth and plausible-looking results. Use field measurements to decide whether the assumption holds instead of relying only on the appearance of a simulation graph.

Principle 3: Gas–liquid contacting

Gas hold-up changes mixture density, reaction area and pressure drop. A simplistic homogeneous approximation helps bookkeeping but misses slip and flow-pattern transitions. In practical calculations, αg≈Qg/(Qg+Ql) provides a compact organizing relationship, but its symbols conceal choices about reference conditions and constitutive data. Before interpreting the output, identify the actual state of the fluid and the geometric scale of the selected section. A flow model should satisfy the limiting condition of zero throughput or zero driving force when that is physically relevant. Such checks can expose a factor-of-four friction convention error or a pressure-reference mistake before any large computational exercise.

Process example. Evaluate an air-lift loop before committing to a two-phase pressure-drop design. First state the unknown to be solved and identify which measurements can constrain it. Write the full relation and check units before estimating the dominant terms. Void fraction often must be measured or correlated for a particular regime. If the result is used in a design discussion, accompany it with a sensitivity study rather than an unqualified single-point value.

Counterexample. Assuming both phases move at equal local velocity in a slug-flow reactor can give erroneous residence times. The lesson is that algebraic correctness is not equivalent to physical correctness. A chosen property, correlation or boundary condition can fail while the numerical method continues to produce smooth and plausible-looking results. Use field measurements to decide whether the assumption holds instead of relying only on the appearance of a simulation graph.

Principle 4: Slurry transport and settling

Fine solids may remain suspended at modest speed while coarse dense solids can settle and abrade the line. In practical calculations, vsettling≈(ρp−ρf)gdp²/(18μ) provides a compact organizing relationship, but its symbols conceal choices about reference conditions and constitutive data. Before interpreting the output, identify the actual state of the fluid and the geometric scale of the selected section. A flow model should satisfy the limiting condition of zero throughput or zero driving force when that is physically relevant. Such checks can expose a factor-of-four friction convention error or a pressure-reference mistake before any large computational exercise.

Process example. Screen settling risk in a mineral-processing feed pipe during a low-flow operating period. First state the unknown to be solved and identify which measurements can constrain it. Write the full relation and check units before estimating the dominant terms. Particle interactions and hindered settling invalidate the dilute-Stokes approximation. If the result is used in a design discussion, accompany it with a sensitivity study rather than an unqualified single-point value.

Counterexample. Applying Stokes settling beyond small particle Reynolds number is a common serious mistake. The lesson is that algebraic correctness is not equivalent to physical correctness. A chosen property, correlation or boundary condition can fail while the numerical method continues to produce smooth and plausible-looking results. Use field measurements to decide whether the assumption holds instead of relying only on the appearance of a simulation graph.

An integrated diagnostic walkthrough

Imagine that a production line delivers noticeably less liquid than the specified target. Check whether mass conservation closes around the known boundary and whether the meter reading refers to actual operating density. Next compare the system's elevation head with the estimated line losses, and locate the pressure measurements relative to fittings or flow disturbances. A pump and a valve affect the same operating point, so they cannot be diagnosed independently without tracing the entire system curve. During a shutdown or restart the steady-state model may not apply at all.

Now consider a second observation: the same line operates normally when cold but shows a deficit when warm. If the liquid is Newtonian and its viscosity decreases with temperature, the hydraulic interpretation may be very different from a liquid whose vapor pressure rises enough to reduce pump suction margin. Identify both mechanisms before choosing one. If the fluid contains suspended solids, particle settling and changes in concentration may dominate. These hypotheses require independent observations, not simply adjusting one friction coefficient until the curve passes through a measured point.

A useful sensitivity calculation perturbs at least three distinct input categories: a material property, a geometry, and an imposed operating condition. Compare the effect on predicted flow or energy in common units. A poorly identified parameter can be acceptable in an early screening model if the final decision is insensitive to its full plausible range. Conversely, a parameter that strongly affects a safety or equipment boundary must be measured more carefully before that decision is made.

Unit synthesis and problem-solving sequence

Create a schematic of the process, define the control surfaces, and label pressure bases, elevations, diameters, materials and flow directions. Select the applicable relationships, including the distinctions between conservation laws and empirical closures. Convert all quantities to consistent units. Compute a baseline solution, run two boundary checks, and evaluate sensitivity to the uncertain parameters. Finally compare the result with measurements or a known reference calculation. A mismatch should trigger investigation of missing physics, not immediate adjustment of an arbitrary calibration factor.

Four lessons and their models

• [Packed-bed friction](../lessons/unit-06-lesson-01.html)
• [Fluidization onset](../lessons/unit-06-lesson-02.html)
• [Gas–liquid contacting](../lessons/unit-06-lesson-03.html)
• [Slurry transport and settling](../lessons/unit-06-lesson-04.html)

Unit reflection

For each of the four topics, describe an operating regime in which the indicated equation is credible and another where it is not. Which two measurements would most efficiently detect that failure? Explain how you would separate a process change from an instrument problem. Finish with a one-page engineering statement: list assumptions, output units, uncertain inputs, and what additional evidence is required for any industrial design use.

Unit 6: Multiphase flow — Lesson 1: Packed-bed friction

Study position: Packed beds; fluidization; gas-liquid flow; slurries. Model relationship: Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp). Associated laboratory: [Ergun fixed-bed pressure drop](../labs/lab-15.html).

Why this phenomenon matters

Ergun combines a viscous term with a kinetic term. Superficial velocity uses empty vessel cross-section, not the void area. The first decision is whether the quantities in the chosen model are measurable in the actual operation. In estimate pressure required to circulate process gas through a catalyst bed., the equation is useful only after the student writes down the fluid composition, temperature, pressure basis, geometry and direction of every transfer. This prevents a number copied from a property chart from masquerading as a defensible process estimate.

A process engineer should explicitly state which unknown is being solved. For this lesson the core governing relationship is Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp). Rearranging such a relationship does not create new physical information: a pressure-loss prediction still needs a friction model, while a rheology equation still needs measured parameters. The distinction between known geometry, reported operating conditions and fitted material coefficients is fundamental to reliable modeling.

Physical foundations and interpretation

Several industrial contactors exchange momentum between phases. Packed-bed resistance involves a viscous contribution linear in superficial speed and an inertial contribution quadratic in speed. A bed begins to fluidize when pressure drop supports the effective weight of solids; beyond that point the settled-bed Ergun equation is not sufficient by itself. Gas–liquid and slurry correlations are regime-dependent and need experimental verification.

Ergun combines a viscous term with a kinetic term. Superficial velocity uses empty vessel cross-section, not the void area. This statement contains a model, not merely an algebraic trick. Sketch a control volume and label its state variables. Ask whether the relation is local or integrated, whether it is static or dynamic, and whether its coefficients vary through the apparatus. A highly polished result is not more credible if the coefficients were measured at another temperature or process regime.

The effect is important in a routine operating audit. The dominant quantity may change when the characteristic flow, wall condition or operating pressure changes. Consequently, numerical sensitivity should form part of the interpretation rather than being treated as a decorative plot. A successful investigation reports a trend, a dimensionless diagnostic where possible, and the limits of extrapolation.

Equation, quantities, and a derivation strategy

Write Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp) and first identify every dimensional quantity. For an equation obtained by conservation, define a clear system boundary and choose the sign conventions before substituting data. For a constitutive relation, find the regime and the data used to calibrate the parameters. Then test a limiting case: zero flow should remove a passive friction loss; zero gravitational height should remove the static elevation contribution; and a smooth pipe should not be assigned a large rough-wall contribution without evidence.

Units must remain consistent in any intermediate stage. Typical process values are measured in litres per second, millimetres, kilopascals and degrees Celsius, whereas the elementary equations often require cubic metres per second, metres, pascals and kelvins. Keep the input-unit conversions visible in the calculation. A line of carefully checked units often detects an error more effectively than several additional significant figures.

Common error and correction: Inserting the interstitial speed in a superficial-speed correlation distorts both contributions. Correct this by replacing the faulty assumption with an explicit measured quantity or a justified correlation. Particle shape, wall effects and channeling restrict accuracy. State this boundary of validity in any design memo, because the limiting condition is often exactly where the operational risk becomes important.

Worked numerical investigations

The cases below exercise the related laboratory's independent mathematical model: Ergun fixed-bed pressure drop. They illustrate how the numerical process behaves under changed conditions. They are related explorations, not a claim that the laboratory formula is identical to Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp). Each case specifies a manipulated input and holds the other controls at their default values. For a full professional calculation, use documented data and a suitable design correlation.

Investigation 1: Planned equipment retrofit

Set superficial velocity = 0.144 m/s while holding the remaining inputs at the values listed in laboratory 15. The calculated response is 158.103 kPa from Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)). Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this planned equipment retrofit, explain whether bed pressure loss is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, assumes waterlike density/viscosity and a fixed bed of representative spherical particles. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 2: Routine operating audit

Set superficial velocity = 0.3386 m/s while holding the remaining inputs at the values listed in laboratory 15. The calculated response is 822.213 kPa from Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)). Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this routine operating audit, explain whether bed pressure loss is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, assumes waterlike density/viscosity and a fixed bed of representative spherical particles. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 3: Unexpected off-design event

Set superficial velocity = 0.5471 m/s while holding the remaining inputs at the values listed in laboratory 15. The calculated response is 2108.32 kPa from Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)). Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this unexpected off-design event, explain whether bed pressure loss is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, assumes waterlike density/viscosity and a fixed bed of representative spherical particles. A credible report will distinguish a prediction from a calibrated measurement.

Measurement design and evidence

Start by listing the relevant field instruments. Temperature and pressure transmitters locate the fluid state, volumetric or mass-flow meters define transfer, and geometric measurements establish areas and lengths. If packed-bed friction influences the predicted outcome, choose instrument locations carefully: a pressure tapping beside an elbow can register a different condition from a tap in a straight developed section. Record which readings are simultaneous; combining unmatched operating points may look mathematically consistent while describing no actual condition.

Uncertainty is not only sensor tolerance. An assumed roughness, uncertain composition, partially closed bypass, transient load or unrecognized entrained gas can dominate the error. First make a sensitivity table by perturbing the input most directly tied to Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp). Next decide if an additional instrument would distinguish two competing explanations. A repeated measurement tests precision; an independent instrument or a separate physical law helps test systematic error.

Failure analysis and boundary checks

Inserting the interstitial speed in a superficial-speed correlation distorts both contributions. This error would be especially consequential in estimate pressure required to circulate process gas through a catalyst bed. because the calculation might appear to converge while its physical interpretation is invalid. If a flow estimate becomes negative, a passive loss contributes energy, or a predicted device operates outside its curve, stop and audit signs and model validity. Plausibility must be tested before a software output is trusted. The important diagnostic for this lesson is sensitivity to material properties.

To investigate a discrepancy, separate property uncertainty from geometric uncertainty and from instrument bias. Recompute the predicted result at the maximum and minimum plausible properties. Determine whether this range spans the observation. If it does, the mismatch may be underdetermined; if it does not, missing physics or a wrong boundary condition becomes more likely. In a real plant, never alter safety-related constraints to force agreement with a teaching model.

Exercises — questions and explanation prompts

1. Derive the units of Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp) and show each conversion required for a plant datasheet reporting mixed engineering units.
2. Identify the operating regime in which packed-bed friction can be simplified. State at least two reasons why the simplification might fail.
3. Rework Investigation 2 with the primary input reduced by 10%. Calculate the percentage response and explain whether it is proportional.
4. Design a small-scale measurement plan that could falsify the assumed relationship. Identify one independent variable and one confounding variable.
5. Explain how you would document estimate pressure required to circulate process gas through a catalyst bed. in a calculation note so another engineer could reconstruct every numerical assumption.

Answer guidance and self-check

An adequate solution must include the model, an SI-unit conversion, a physical explanation and a limitation. The predicted result for Investigation 2 is 822.213 kPa. Numerical agreement to a few decimals is not enough: explain why it is possible to use Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)) here, and why that model may need replacement for a different fluid, scale or operating range. The correct response to the open questions depends on explicit experimental choices and must be defended with measurements.

Unit 6: Multiphase flow — Lesson 2: Fluidization onset

Study position: Packed beds; fluidization; gas-liquid flow; slurries. Model relationship: Δpbed≈(ρs−ρf)(1−ε)Lg. Associated laboratory: [Minimum fluidization screening](../labs/lab-16.html).

Why this phenomenon matters

Incipient fluidization is approximately the point where fluid drag supports the apparent weight of bed solids. The first decision is whether the quantities in the chosen model are measurable in the actual operation. In predict minimum airflow needed to suspend a granular catalyst in a small test column., the equation is useful only after the student writes down the fluid composition, temperature, pressure basis, geometry and direction of every transfer. This prevents a number copied from a property chart from masquerading as a defensible process estimate.

A process engineer should explicitly state which unknown is being solved. For this lesson the core governing relationship is Δpbed≈(ρs−ρf)(1−ε)Lg. Rearranging such a relationship does not create new physical information: a pressure-loss prediction still needs a friction model, while a rheology equation still needs measured parameters. The distinction between known geometry, reported operating conditions and fitted material coefficients is fundamental to reliable modeling.

Physical foundations and interpretation

Several industrial contactors exchange momentum between phases. Packed-bed resistance involves a viscous contribution linear in superficial speed and an inertial contribution quadratic in speed. A bed begins to fluidize when pressure drop supports the effective weight of solids; beyond that point the settled-bed Ergun equation is not sufficient by itself. Gas–liquid and slurry correlations are regime-dependent and need experimental verification.

Incipient fluidization is approximately the point where fluid drag supports the apparent weight of bed solids. This statement contains a model, not merely an algebraic trick. Sketch a control volume and label its state variables. Ask whether the relation is local or integrated, whether it is static or dynamic, and whether its coefficients vary through the apparatus. A highly polished result is not more credible if the coefficients were measured at another temperature or process regime.

The effect is important in a unexpected off-design event. The dominant quantity may change when the characteristic flow, wall condition or operating pressure changes. Consequently, numerical sensitivity should form part of the interpretation rather than being treated as a decorative plot. A successful investigation reports a trend, a dimensionless diagnostic where possible, and the limits of extrapolation.

Equation, quantities, and a derivation strategy

Write Δpbed≈(ρs−ρf)(1−ε)Lg and first identify every dimensional quantity. For an equation obtained by conservation, define a clear system boundary and choose the sign conventions before substituting data. For a constitutive relation, find the regime and the data used to calibrate the parameters. Then test a limiting case: zero flow should remove a passive friction loss; zero gravitational height should remove the static elevation contribution; and a smooth pipe should not be assigned a large rough-wall contribution without evidence.

Units must remain consistent in any intermediate stage. Typical process values are measured in litres per second, millimetres, kilopascals and degrees Celsius, whereas the elementary equations often require cubic metres per second, metres, pascals and kelvins. Keep the input-unit conversions visible in the calculation. A line of carefully checked units often detects an error more effectively than several additional significant figures.

Common error and correction: The pressure drop does not keep increasing according to the fixed-bed Ergun equation after a bed expands. Correct this by replacing the faulty assumption with an explicit measured quantity or a justified correlation. Cohesion, irregular particles and bubble formation require more detailed correlations. State this boundary of validity in any design memo, because the limiting condition is often exactly where the operational risk becomes important.

Worked numerical investigations

The cases below exercise the related laboratory's independent mathematical model: Minimum fluidization screening. They illustrate how the numerical process behaves under changed conditions. They are related explorations, not a claim that the laboratory formula is identical to Δpbed≈(ρs−ρf)(1−ε)Lg. Each case specifies a manipulated input and holds the other controls at their default values. For a full professional calculation, use documented data and a suitable design correlation.

Investigation 1: Routine operating audit

Set particle size = 1.16 mm while holding the remaining inputs at the values listed in laboratory 16. The calculated response is 0.730319 m/s from Ergun drag balances effective bed weight. Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this routine operating audit, explain whether estimated minimum fluidization velocity is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, gas-phase screening at ρ=1.2 kg/m³ and μ=1.8×10⁻⁵ Pa·s; not a substitute for fluidization experiments. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 2: Unexpected off-design event

Set particle size = 2.504 mm while holding the remaining inputs at the values listed in laboratory 16. The calculated response is 1.32993 m/s from Ergun drag balances effective bed weight. Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this unexpected off-design event, explain whether estimated minimum fluidization velocity is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, gas-phase screening at ρ=1.2 kg/m³ and μ=1.8×10⁻⁵ Pa·s; not a substitute for fluidization experiments. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 3: Scale-up screening

Set particle size = 3.944 mm while holding the remaining inputs at the values listed in laboratory 16. The calculated response is 1.75442 m/s from Ergun drag balances effective bed weight. Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this scale-up screening, explain whether estimated minimum fluidization velocity is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, gas-phase screening at ρ=1.2 kg/m³ and μ=1.8×10⁻⁵ Pa·s; not a substitute for fluidization experiments. A credible report will distinguish a prediction from a calibrated measurement.

Measurement design and evidence

Start by listing the relevant field instruments. Temperature and pressure transmitters locate the fluid state, volumetric or mass-flow meters define transfer, and geometric measurements establish areas and lengths. If fluidization onset influences the predicted outcome, choose instrument locations carefully: a pressure tapping beside an elbow can register a different condition from a tap in a straight developed section. Record which readings are simultaneous; combining unmatched operating points may look mathematically consistent while describing no actual condition.

Uncertainty is not only sensor tolerance. An assumed roughness, uncertain composition, partially closed bypass, transient load or unrecognized entrained gas can dominate the error. First make a sensitivity table by perturbing the input most directly tied to Δpbed≈(ρs−ρf)(1−ε)Lg. Next decide if an additional instrument would distinguish two competing explanations. A repeated measurement tests precision; an independent instrument or a separate physical law helps test systematic error.

Failure analysis and boundary checks

The pressure drop does not keep increasing according to the fixed-bed Ergun equation after a bed expands. This error would be especially consequential in predict minimum airflow needed to suspend a granular catalyst in a small test column. because the calculation might appear to converge while its physical interpretation is invalid. If a flow estimate becomes negative, a passive loss contributes energy, or a predicted device operates outside its curve, stop and audit signs and model validity. Plausibility must be tested before a software output is trusted. The important diagnostic for this lesson is instrument calibration.

To investigate a discrepancy, separate property uncertainty from geometric uncertainty and from instrument bias. Recompute the predicted result at the maximum and minimum plausible properties. Determine whether this range spans the observation. If it does, the mismatch may be underdetermined; if it does not, missing physics or a wrong boundary condition becomes more likely. In a real plant, never alter safety-related constraints to force agreement with a teaching model.

Exercises — questions and explanation prompts

1. Derive the units of Δpbed≈(ρs−ρf)(1−ε)Lg and show each conversion required for a plant datasheet reporting mixed engineering units.
2. Identify the operating regime in which fluidization onset can be simplified. State at least two reasons why the simplification might fail.
3. Rework Investigation 2 with the primary input reduced by 10%. Calculate the percentage response and explain whether it is proportional.
4. Design a small-scale measurement plan that could falsify the assumed relationship. Identify one independent variable and one confounding variable.
5. Explain how you would document predict minimum airflow needed to suspend a granular catalyst in a small test column. in a calculation note so another engineer could reconstruct every numerical assumption.

Answer guidance and self-check

An adequate solution must include the model, an SI-unit conversion, a physical explanation and a limitation. The predicted result for Investigation 2 is 1.32993 m/s. Numerical agreement to a few decimals is not enough: explain why it is possible to use Ergun drag balances effective bed weight here, and why that model may need replacement for a different fluid, scale or operating range. The correct response to the open questions depends on explicit experimental choices and must be defended with measurements.

Unit 6: Multiphase flow — Lesson 3: Gas–liquid contacting

Study position: Packed beds; fluidization; gas-liquid flow; slurries. Model relationship: αg≈Qg/(Qg+Ql). Associated laboratory: [Ergun fixed-bed pressure drop](../labs/lab-15.html).

Why this phenomenon matters

Gas hold-up changes mixture density, reaction area and pressure drop. A simplistic homogeneous approximation helps bookkeeping but misses slip and flow-pattern transitions. The first decision is whether the quantities in the chosen model are measurable in the actual operation. In evaluate an air-lift loop before committing to a two-phase pressure-drop design., the equation is useful only after the student writes down the fluid composition, temperature, pressure basis, geometry and direction of every transfer. This prevents a number copied from a property chart from masquerading as a defensible process estimate.

A process engineer should explicitly state which unknown is being solved. For this lesson the core governing relationship is αg≈Qg/(Qg+Ql). Rearranging such a relationship does not create new physical information: a pressure-loss prediction still needs a friction model, while a rheology equation still needs measured parameters. The distinction between known geometry, reported operating conditions and fitted material coefficients is fundamental to reliable modeling.

Physical foundations and interpretation

Several industrial contactors exchange momentum between phases. Packed-bed resistance involves a viscous contribution linear in superficial speed and an inertial contribution quadratic in speed. A bed begins to fluidize when pressure drop supports the effective weight of solids; beyond that point the settled-bed Ergun equation is not sufficient by itself. Gas–liquid and slurry correlations are regime-dependent and need experimental verification.

Gas hold-up changes mixture density, reaction area and pressure drop. A simplistic homogeneous approximation helps bookkeeping but misses slip and flow-pattern transitions. This statement contains a model, not merely an algebraic trick. Sketch a control volume and label its state variables. Ask whether the relation is local or integrated, whether it is static or dynamic, and whether its coefficients vary through the apparatus. A highly polished result is not more credible if the coefficients were measured at another temperature or process regime.

The effect is important in a scale-up screening. The dominant quantity may change when the characteristic flow, wall condition or operating pressure changes. Consequently, numerical sensitivity should form part of the interpretation rather than being treated as a decorative plot. A successful investigation reports a trend, a dimensionless diagnostic where possible, and the limits of extrapolation.

Equation, quantities, and a derivation strategy

Write αg≈Qg/(Qg+Ql) and first identify every dimensional quantity. For an equation obtained by conservation, define a clear system boundary and choose the sign conventions before substituting data. For a constitutive relation, find the regime and the data used to calibrate the parameters. Then test a limiting case: zero flow should remove a passive friction loss; zero gravitational height should remove the static elevation contribution; and a smooth pipe should not be assigned a large rough-wall contribution without evidence.

Units must remain consistent in any intermediate stage. Typical process values are measured in litres per second, millimetres, kilopascals and degrees Celsius, whereas the elementary equations often require cubic metres per second, metres, pascals and kelvins. Keep the input-unit conversions visible in the calculation. A line of carefully checked units often detects an error more effectively than several additional significant figures.

Common error and correction: Assuming both phases move at equal local velocity in a slug-flow reactor can give erroneous residence times. Correct this by replacing the faulty assumption with an explicit measured quantity or a justified correlation. Void fraction often must be measured or correlated for a particular regime. State this boundary of validity in any design memo, because the limiting condition is often exactly where the operational risk becomes important.

Worked numerical investigations

The cases below exercise the related laboratory's independent mathematical model: Ergun fixed-bed pressure drop. They illustrate how the numerical process behaves under changed conditions. They are related explorations, not a claim that the laboratory formula is identical to αg≈Qg/(Qg+Ql). Each case specifies a manipulated input and holds the other controls at their default values. For a full professional calculation, use documented data and a suitable design correlation.

Investigation 1: Unexpected off-design event

Set superficial velocity = 0.144 m/s while holding the remaining inputs at the values listed in laboratory 15. The calculated response is 158.103 kPa from Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)). Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this unexpected off-design event, explain whether bed pressure loss is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, assumes waterlike density/viscosity and a fixed bed of representative spherical particles. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 2: Scale-up screening

Set superficial velocity = 0.3386 m/s while holding the remaining inputs at the values listed in laboratory 15. The calculated response is 822.213 kPa from Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)). Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this scale-up screening, explain whether bed pressure loss is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, assumes waterlike density/viscosity and a fixed bed of representative spherical particles. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 3: Laboratory demonstrator

Set superficial velocity = 0.5471 m/s while holding the remaining inputs at the values listed in laboratory 15. The calculated response is 2108.32 kPa from Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)). Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this laboratory demonstrator, explain whether bed pressure loss is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, assumes waterlike density/viscosity and a fixed bed of representative spherical particles. A credible report will distinguish a prediction from a calibrated measurement.

Measurement design and evidence

Start by listing the relevant field instruments. Temperature and pressure transmitters locate the fluid state, volumetric or mass-flow meters define transfer, and geometric measurements establish areas and lengths. If gas–liquid contacting influences the predicted outcome, choose instrument locations carefully: a pressure tapping beside an elbow can register a different condition from a tap in a straight developed section. Record which readings are simultaneous; combining unmatched operating points may look mathematically consistent while describing no actual condition.

Uncertainty is not only sensor tolerance. An assumed roughness, uncertain composition, partially closed bypass, transient load or unrecognized entrained gas can dominate the error. First make a sensitivity table by perturbing the input most directly tied to αg≈Qg/(Qg+Ql). Next decide if an additional instrument would distinguish two competing explanations. A repeated measurement tests precision; an independent instrument or a separate physical law helps test systematic error.

Failure analysis and boundary checks

Assuming both phases move at equal local velocity in a slug-flow reactor can give erroneous residence times. This error would be especially consequential in evaluate an air-lift loop before committing to a two-phase pressure-drop design. because the calculation might appear to converge while its physical interpretation is invalid. If a flow estimate becomes negative, a passive loss contributes energy, or a predicted device operates outside its curve, stop and audit signs and model validity. Plausibility must be tested before a software output is trusted. The important diagnostic for this lesson is plausibility in limiting conditions.

To investigate a discrepancy, separate property uncertainty from geometric uncertainty and from instrument bias. Recompute the predicted result at the maximum and minimum plausible properties. Determine whether this range spans the observation. If it does, the mismatch may be underdetermined; if it does not, missing physics or a wrong boundary condition becomes more likely. In a real plant, never alter safety-related constraints to force agreement with a teaching model.

Exercises — questions and explanation prompts

1. Derive the units of αg≈Qg/(Qg+Ql) and show each conversion required for a plant datasheet reporting mixed engineering units.
2. Identify the operating regime in which gas–liquid contacting can be simplified. State at least two reasons why the simplification might fail.
3. Rework Investigation 2 with the primary input reduced by 10%. Calculate the percentage response and explain whether it is proportional.
4. Design a small-scale measurement plan that could falsify the assumed relationship. Identify one independent variable and one confounding variable.
5. Explain how you would document evaluate an air-lift loop before committing to a two-phase pressure-drop design. in a calculation note so another engineer could reconstruct every numerical assumption.

Answer guidance and self-check

An adequate solution must include the model, an SI-unit conversion, a physical explanation and a limitation. The predicted result for Investigation 2 is 822.213 kPa. Numerical agreement to a few decimals is not enough: explain why it is possible to use Δp=L(150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp)) here, and why that model may need replacement for a different fluid, scale or operating range. The correct response to the open questions depends on explicit experimental choices and must be defended with measurements.

Unit 6: Multiphase flow — Lesson 4: Slurry transport and settling

Study position: Packed beds; fluidization; gas-liquid flow; slurries. Model relationship: vsettling≈(ρp−ρf)gdp²/(18μ). Associated laboratory: [Minimum fluidization screening](../labs/lab-16.html).

Why this phenomenon matters

Fine solids may remain suspended at modest speed while coarse dense solids can settle and abrade the line. The first decision is whether the quantities in the chosen model are measurable in the actual operation. In screen settling risk in a mineral-processing feed pipe during a low-flow operating period., the equation is useful only after the student writes down the fluid composition, temperature, pressure basis, geometry and direction of every transfer. This prevents a number copied from a property chart from masquerading as a defensible process estimate.

A process engineer should explicitly state which unknown is being solved. For this lesson the core governing relationship is vsettling≈(ρp−ρf)gdp²/(18μ). Rearranging such a relationship does not create new physical information: a pressure-loss prediction still needs a friction model, while a rheology equation still needs measured parameters. The distinction between known geometry, reported operating conditions and fitted material coefficients is fundamental to reliable modeling.

Physical foundations and interpretation

Several industrial contactors exchange momentum between phases. Packed-bed resistance involves a viscous contribution linear in superficial speed and an inertial contribution quadratic in speed. A bed begins to fluidize when pressure drop supports the effective weight of solids; beyond that point the settled-bed Ergun equation is not sufficient by itself. Gas–liquid and slurry correlations are regime-dependent and need experimental verification.

Fine solids may remain suspended at modest speed while coarse dense solids can settle and abrade the line. This statement contains a model, not merely an algebraic trick. Sketch a control volume and label its state variables. Ask whether the relation is local or integrated, whether it is static or dynamic, and whether its coefficients vary through the apparatus. A highly polished result is not more credible if the coefficients were measured at another temperature or process regime.

The effect is important in a laboratory demonstrator. The dominant quantity may change when the characteristic flow, wall condition or operating pressure changes. Consequently, numerical sensitivity should form part of the interpretation rather than being treated as a decorative plot. A successful investigation reports a trend, a dimensionless diagnostic where possible, and the limits of extrapolation.

Equation, quantities, and a derivation strategy

Write vsettling≈(ρp−ρf)gdp²/(18μ) and first identify every dimensional quantity. For an equation obtained by conservation, define a clear system boundary and choose the sign conventions before substituting data. For a constitutive relation, find the regime and the data used to calibrate the parameters. Then test a limiting case: zero flow should remove a passive friction loss; zero gravitational height should remove the static elevation contribution; and a smooth pipe should not be assigned a large rough-wall contribution without evidence.

Units must remain consistent in any intermediate stage. Typical process values are measured in litres per second, millimetres, kilopascals and degrees Celsius, whereas the elementary equations often require cubic metres per second, metres, pascals and kelvins. Keep the input-unit conversions visible in the calculation. A line of carefully checked units often detects an error more effectively than several additional significant figures.

Common error and correction: Applying Stokes settling beyond small particle Reynolds number is a common serious mistake. Correct this by replacing the faulty assumption with an explicit measured quantity or a justified correlation. Particle interactions and hindered settling invalidate the dilute-Stokes approximation. State this boundary of validity in any design memo, because the limiting condition is often exactly where the operational risk becomes important.

Worked numerical investigations

The cases below exercise the related laboratory's independent mathematical model: Minimum fluidization screening. They illustrate how the numerical process behaves under changed conditions. They are related explorations, not a claim that the laboratory formula is identical to vsettling≈(ρp−ρf)gdp²/(18μ). Each case specifies a manipulated input and holds the other controls at their default values. For a full professional calculation, use documented data and a suitable design correlation.

Investigation 1: Scale-up screening

Set particle size = 1.16 mm while holding the remaining inputs at the values listed in laboratory 16. The calculated response is 0.730319 m/s from Ergun drag balances effective bed weight. Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this scale-up screening, explain whether estimated minimum fluidization velocity is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, gas-phase screening at ρ=1.2 kg/m³ and μ=1.8×10⁻⁵ Pa·s; not a substitute for fluidization experiments. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 2: Laboratory demonstrator

Set particle size = 2.504 mm while holding the remaining inputs at the values listed in laboratory 16. The calculated response is 1.32993 m/s from Ergun drag balances effective bed weight. Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this laboratory demonstrator, explain whether estimated minimum fluidization velocity is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, gas-phase screening at ρ=1.2 kg/m³ and μ=1.8×10⁻⁵ Pa·s; not a substitute for fluidization experiments. A credible report will distinguish a prediction from a calibrated measurement.

Investigation 3: Pilot-plant commissioning

Set particle size = 3.944 mm while holding the remaining inputs at the values listed in laboratory 16. The calculated response is 1.75442 m/s from Ergun drag balances effective bed weight. Convert the displayed input into coherent SI units before independently reproducing the result. If the result is unexpectedly large or small, estimate the scale using a rough hand calculation and inspect each power of length or flow in the equation.

For this pilot-plant commissioning, explain whether estimated minimum fluidization velocity is the decisive metric or a supporting intermediate quantity. A change in this one result might imply a different operating point or energy demand, but it does not by itself approve a pump, pipe or valve. Write down which additional measurement would make the inference stronger. In particular, gas-phase screening at ρ=1.2 kg/m³ and μ=1.8×10⁻⁵ Pa·s; not a substitute for fluidization experiments. A credible report will distinguish a prediction from a calibrated measurement.

Measurement design and evidence

Start by listing the relevant field instruments. Temperature and pressure transmitters locate the fluid state, volumetric or mass-flow meters define transfer, and geometric measurements establish areas and lengths. If slurry transport and settling influences the predicted outcome, choose instrument locations carefully: a pressure tapping beside an elbow can register a different condition from a tap in a straight developed section. Record which readings are simultaneous; combining unmatched operating points may look mathematically consistent while describing no actual condition.

Uncertainty is not only sensor tolerance. An assumed roughness, uncertain composition, partially closed bypass, transient load or unrecognized entrained gas can dominate the error. First make a sensitivity table by perturbing the input most directly tied to vsettling≈(ρp−ρf)gdp²/(18μ). Next decide if an additional instrument would distinguish two competing explanations. A repeated measurement tests precision; an independent instrument or a separate physical law helps test systematic error.

Failure analysis and boundary checks

Applying Stokes settling beyond small particle Reynolds number is a common serious mistake. This error would be especially consequential in screen settling risk in a mineral-processing feed pipe during a low-flow operating period. because the calculation might appear to converge while its physical interpretation is invalid. If a flow estimate becomes negative, a passive loss contributes energy, or a predicted device operates outside its curve, stop and audit signs and model validity. Plausibility must be tested before a software output is trusted. The important diagnostic for this lesson is agreement with a separate empirical measurement.

To investigate a discrepancy, separate property uncertainty from geometric uncertainty and from instrument bias. Recompute the predicted result at the maximum and minimum plausible properties. Determine whether this range spans the observation. If it does, the mismatch may be underdetermined; if it does not, missing physics or a wrong boundary condition becomes more likely. In a real plant, never alter safety-related constraints to force agreement with a teaching model.

Exercises — questions and explanation prompts

1. Derive the units of vsettling≈(ρp−ρf)gdp²/(18μ) and show each conversion required for a plant datasheet reporting mixed engineering units.
2. Identify the operating regime in which slurry transport and settling can be simplified. State at least two reasons why the simplification might fail.
3. Rework Investigation 2 with the primary input reduced by 10%. Calculate the percentage response and explain whether it is proportional.
4. Design a small-scale measurement plan that could falsify the assumed relationship. Identify one independent variable and one confounding variable.
5. Explain how you would document screen settling risk in a mineral-processing feed pipe during a low-flow operating period. in a calculation note so another engineer could reconstruct every numerical assumption.

Answer guidance and self-check

An adequate solution must include the model, an SI-unit conversion, a physical explanation and a limitation. The predicted result for Investigation 2 is 1.32993 m/s. Numerical agreement to a few decimals is not enough: explain why it is possible to use Ergun drag balances effective bed weight here, and why that model may need replacement for a different fluid, scale or operating range. The correct response to the open questions depends on explicit experimental choices and must be defended with measurements.

Unit 6 applied project: Multiphase flow

Design brief

A medium-size chemical plant is assessing a fluid transfer or processing subsystem. Your assigned topic is multiphase flow. The commissioning dossier contains a process sketch, a partial set of flow and pressure readings, and uncertainty in at least one fluid property. Construct a transparent calculation that can be checked independently. The aim is to evaluate available evidence, not to certify equipment for a hazardous duty.

Specific project route

Several industrial contactors exchange momentum between phases. Packed-bed resistance involves a viscous contribution linear in superficial speed and an inertial contribution quadratic in speed. A bed begins to fluidize when pressure drop supports the effective weight of solids; beyond that point the settled-bed Ergun equation is not sufficient by itself. Gas–liquid and slurry correlations are regime-dependent and need experimental verification.

Build a measurement plan with a process flow diagram and separate mass, energy, and equipment assumptions. Choose two laboratory models related to this unit, identify which governing equations overlap and which assumptions conflict, and reconcile them by writing a joint list of state variables. Compare two plausible operating points and explain which one you would investigate first. If the calculations cannot determine a unique operating point, show explicitly which additional measured condition is missing.

Required outputs

1. A process sketch showing node elevations, flow directions, measurement locations and equipment labels.
2. A one-page derivation, with dimensions checked and at least one explicitly computed limiting case.
3. A CSV dataset generated by changing an appropriate laboratory model input across its allowed range. Include plots with physical units.
4. A sensitivity analysis that changes one material parameter, one boundary condition and one equipment or geometry parameter independently.
5. A decision note that separates validated physical relationships, assumed correlations and unresolved uncertainty.

Diagnostic questions

What observed trend would disprove the selected model? How would your process react if temperature increased while a valve became partially blocked? Could the same output change be explained by a sensor error? When would running an experiment on the real equipment be unsafe or unnecessary because the model has already violated a pressure or flow limit?

Assessment: Technical clarity 30%; defensible equations 25%; documented calculations 20%; sensitivity and verification 15%; communication and honest scope limits 10%.

Laboratory 11: Pump and system intersection

Purpose and governing physics

Primary observation: Operating flow (L/s). Equation: Q*=sqrt((H0−Hstatic)/(a+k)).

Quadratic pump and system curves are illustrative. A nonpositive intersection means this pump cannot meet the static lift. The model uses explicit units and a clearly selected control variable, rather than displaying an arbitrary animation. Examine it as a hypothesis: it is an engineering relationship with assumptions, not a guarantee for every real material or installation.

Before interacting

1. Identify the physically independent variable: static head in m. What property or device controls it in the actual plant?
2. Confirm that Q*=sqrt((H0−Hstatic)/(a+k)) predicts a dimensionally consistent quantity. Identify any constants or coefficients that require measurement rather than theory.
3. Predict qualitatively whether the displayed operating flow will increase, decrease, or reach a limiting value. Record the prediction before changing the slider.

Worked baseline

With static head = 10 m, the displayed model predicts 9.51662 L/s using the other control values shown below. This is a numerical consequence of the simplified equation, not an empirical observation.

Sensitivity experiment

Hold all other controls fixed and examine the output at approximately 15% and 80% of the horizontal-axis range. The first case gives 9.93374 L/s, whereas the second gives 7.21633 L/s. The difference demonstrates sensitivity of the equation over this range. Check for nonlinear curvature rather than inferring a constant slope from one pair of points.

Now change exactly one secondary control, if available. Predict the effect first and then observe how the whole curve changes. Distinguish changes that shift an entire graph from changes that alter its slope or create a limiting value. Export the CSV to calculate a finite-difference sensitivity from two nearby points; report the input perturbation and measurement units.

Interpretation, failure modes and extension

Quadratic pump and system curves are illustrative. A nonpositive intersection means this pump cannot meet the static lift. A useful result is not only a number but a statement of the domain in which that number should be believed. If the model predicts negative flow, unattainable efficiency, or implausibly large pressure, check parameter choices, the sign convention and whether the equation has been extrapolated beyond its assumptions. Do not use the apparent precision of the displayed decimals as an estimate of actual physical accuracy.

For an engineering extension, identify one measured input and one unknown input, propose an instrument and a reasonable calibration experiment, and specify how input uncertainty would propagate into operating flow. Compare the ideal trend with what plant data could look like if temperature changed, a valve fouled, or a sensor drifted. A diagram on the interactive page is schematic and not to scale.

Investigation questions

1. Which assumption would first fail if this process were scaled to a plant with a tenfold larger characteristic flow or vessel size?
2. If static head increases but the observed operating flow moves in the opposite direction to the model, identify at least two plausible missing physical effects.
3. Derive the dimensions of Q*=sqrt((H0−Hstatic)/(a+k)) and explain why reporting the wrong viscosity, pressure basis or density could reverse a process decision.
4. Describe a practical experiment that could distinguish model error from instrument error. State the minimum measurements required.

[Open interactive model](lab-11.html)

Laboratory 12: Pump affinity with speed

Purpose and governing physics

Primary observation: Scaled shaft power (kW). Equation: P=P0(N/N0)³.

Affinity scaling presumes similar operating region and unchanged impeller diameter. The model uses explicit units and a clearly selected control variable, rather than displaying an arbitrary animation. Examine it as a hypothesis: it is an engineering relationship with assumptions, not a guarantee for every real material or installation.

Before interacting

1. Identify the physically independent variable: speed fraction N/N0 in —. What property or device controls it in the actual plant?
2. Confirm that P=P0(N/N0)³ predicts a dimensionally consistent quantity. Identify any constants or coefficients that require measurement rather than theory.
3. Predict qualitatively whether the displayed scaled shaft power will increase, decrease, or reach a limiting value. Record the prediction before changing the slider.

Worked baseline

With speed fraction N/N0 = 0.88 —, the displayed model predicts 8.17766 kW using the other control values shown below. This is a numerical consequence of the simplified equation, not an empirical observation.

Sensitivity experiment

Hold all other controls fixed and examine the output at approximately 15% and 80% of the horizontal-axis range. The first case gives 2.13574 kW, whereas the second gives 13.8915 kW. The difference demonstrates sensitivity of the equation over this range. Check for nonlinear curvature rather than inferring a constant slope from one pair of points.

Now change exactly one secondary control, if available. Predict the effect first and then observe how the whole curve changes. Distinguish changes that shift an entire graph from changes that alter its slope or create a limiting value. Export the CSV to calculate a finite-difference sensitivity from two nearby points; report the input perturbation and measurement units.

Interpretation, failure modes and extension

Affinity scaling presumes similar operating region and unchanged impeller diameter. A useful result is not only a number but a statement of the domain in which that number should be believed. If the model predicts negative flow, unattainable efficiency, or implausibly large pressure, check parameter choices, the sign convention and whether the equation has been extrapolated beyond its assumptions. Do not use the apparent precision of the displayed decimals as an estimate of actual physical accuracy.

For an engineering extension, identify one measured input and one unknown input, propose an instrument and a reasonable calibration experiment, and specify how input uncertainty would propagate into scaled shaft power. Compare the ideal trend with what plant data could look like if temperature changed, a valve fouled, or a sensor drifted. A diagram on the interactive page is schematic and not to scale.

Investigation questions

1. Which assumption would first fail if this process were scaled to a plant with a tenfold larger characteristic flow or vessel size?
2. If speed fraction N/N0 increases but the observed scaled shaft power moves in the opposite direction to the model, identify at least two plausible missing physical effects.
3. Derive the dimensions of P=P0(N/N0)³ and explain why reporting the wrong viscosity, pressure basis or density could reverse a process decision.
4. Describe a practical experiment that could distinguish model error from instrument error. State the minimum measurements required.

[Open interactive model](lab-12.html)

Unit 6: Multiphase flow — assessment

Instructions: Each answer must include a physical explanation and dimensional check. Use the corresponding interactive laboratories for exploration, not as substitute proof. The questions are intended to test derivation, model choice, diagnostic reasoning and quantitative sensitivity.

Topic 1: Packed-bed friction

1. Starting with Δp/L=150μ(1−ε)²u/(ε³dp²)+1.75ρ(1−ε)u²/(ε³dp), define each quantity and derive the expected SI dimensions. If the result is not dimensionally consistent, identify which conversion is likely missing.
2. For the practical setting, estimate pressure required to circulate process gas through a catalyst bed., list the minimum measurements required for a credible baseline calculation. Which input would you vary first in a sensitivity analysis and why?
3. Inserting the interstitial speed in a superficial-speed correlation distorts both contributions. Explain the consequence of this error and design a specific falsification test using two independent observations.
4. The associated laboratory varies superficial velocity (m/s). If that input changes by 20%, will the output change by exactly 20%? Demonstrate using two numerical evaluations; note whether the relationship is linear in the studied interval.

Topic 2: Fluidization onset

1. Starting with Δpbed≈(ρs−ρf)(1−ε)Lg, define each quantity and derive the expected SI dimensions. If the result is not dimensionally consistent, identify which conversion is likely missing.
2. For the practical setting, predict minimum airflow needed to suspend a granular catalyst in a small test column., list the minimum measurements required for a credible baseline calculation. Which input would you vary first in a sensitivity analysis and why?
3. The pressure drop does not keep increasing according to the fixed-bed Ergun equation after a bed expands. Explain the consequence of this error and design a specific falsification test using two independent observations.
4. The associated laboratory varies particle size (mm). If that input changes by 20%, will the output change by exactly 20%? Demonstrate using two numerical evaluations; note whether the relationship is linear in the studied interval.

Topic 3: Gas–liquid contacting

1. Starting with αg≈Qg/(Qg+Ql), define each quantity and derive the expected SI dimensions. If the result is not dimensionally consistent, identify which conversion is likely missing.
2. For the practical setting, evaluate an air-lift loop before committing to a two-phase pressure-drop design., list the minimum measurements required for a credible baseline calculation. Which input would you vary first in a sensitivity analysis and why?
3. Assuming both phases move at equal local velocity in a slug-flow reactor can give erroneous residence times. Explain the consequence of this error and design a specific falsification test using two independent observations.
4. The associated laboratory varies superficial velocity (m/s). If that input changes by 20%, will the output change by exactly 20%? Demonstrate using two numerical evaluations; note whether the relationship is linear in the studied interval.

Topic 4: Slurry transport and settling

1. Starting with vsettling≈(ρp−ρf)gdp²/(18μ), define each quantity and derive the expected SI dimensions. If the result is not dimensionally consistent, identify which conversion is likely missing.
2. For the practical setting, screen settling risk in a mineral-processing feed pipe during a low-flow operating period., list the minimum measurements required for a credible baseline calculation. Which input would you vary first in a sensitivity analysis and why?
3. Applying Stokes settling beyond small particle Reynolds number is a common serious mistake. Explain the consequence of this error and design a specific falsification test using two independent observations.
4. The associated laboratory varies particle size (mm). If that input changes by 20%, will the output change by exactly 20%? Demonstrate using two numerical evaluations; note whether the relationship is linear in the studied interval.

Scoring guidance

Award 30% for correct physical modeling and regimes, 25% for quantitative steps and units, 25% for interpretation and sensitivity, and 20% for validation, evidence and honest acknowledgment of limitations. The open diagnostic questions deliberately admit more than one defensible answer; unsupported claims receive no credit.