DEV Community: NovaSolver

Entropy of Mixing: Why Gases Spread Out and Never Come Back

NovaSolver — Wed, 12 Aug 2026 12:17:02 +0000

Open a bottle of perfume in the corner of a still room and, after a while, you can smell it across the floor. Nobody stirred the air. No fan ran. The scent molecules simply spread until they were everywhere, and they will never, on their own, gather back into the bottle. Run that experiment a billion times and the gas never unmixes. The arrow in that story has a name and a number: the entropy of mixing.

This article explains what entropy of mixing measures, derives the formula for ideal gases, works a clean numerical example, and clears up the misconceptions that surround this deceptively simple result.

Why this calculation matters

Mixing entropy is the quantitative reason that separation costs energy. Pulling nitrogen and oxygen apart from air, stripping carbon dioxide from a flue stream, desalinating seawater, refining a metal from its alloy — every one of these fights against the entropy that was released when the components mixed in the first place. The mixing entropy sets a hard thermodynamic floor on the work any separation process must supply.

It also explains a fact so familiar we forget it needs explaining: why mixtures form spontaneously at all. Two ideal gases at the same temperature and pressure have no energy of interaction to gain or lose by mixing. The drive is purely statistical. Once you can calculate the entropy change, you can compute the Gibbs free energy of mixing, predict whether a blend is stable, and understand why some pairs mix freely while others separate into layers. For chemical engineers, materials scientists, and anyone modelling solutions, this is foundational bookkeeping.

The core method

When two or more different ideal gases mix at constant temperature and pressure, the entropy of the system rises. The increase is the entropy of mixing:

delta_S_mix = -n_total * R * sum(x_i * ln x_i)

Here n_total is the total number of moles, R is the gas constant (8.314 J/mol.K), and x_i is the mole fraction of component i — its share of the total moles. The sum runs over every component in the mixture.

The formula is always positive, and it is worth seeing why. Each mole fraction x_i is between 0 and 1, so its natural logarithm ln x_i is negative. The sum of x_i times ln x_i is therefore negative, and the leading minus sign flips it positive. Entropy always increases on mixing — there is no combination of ideal gases for which this formula returns a negative number.

The physical reading is about counting. Before mixing, each gas occupies its own container; after the partition is removed, every molecule has the whole volume to roam. The number of microscopic arrangements available to the system explodes, and entropy is the logarithm of that count. The mixture is overwhelmingly more probable than the separated state simply because there are vastly more ways to be mixed than to be sorted.

Two companion quantities follow immediately. The Gibbs free energy of mixing is delta_G_mix = -T * delta_S_mix, always negative, which is the formal statement that mixing is spontaneous. And for a given number of components, the entropy of mixing is largest when the mixture is equimolar — equal mole fractions of every species.

A worked example

Mix equal amounts of two different ideal gases at constant temperature and pressure. Each gas contributes 1 mole, so the total is n_total = 2 mol and each mole fraction is x = 0.5.

Step 1 — write the formula for two components.

delta_S_mix = -n_total * R * (x_1*ln x_1 + x_2*ln x_2)

Step 2 — substitute the mole fractions. Both are 0.5, so the bracket has two identical terms:

x_1*ln x_1 + x_2*ln x_2 = 0.5*ln(0.5) + 0.5*ln(0.5) = ln(0.5)

The natural log of 0.5 is about -0.693.

Step 3 — assemble the result.

delta_S_mix = -2 * 8.314 * ln(0.5)
delta_S_mix = -2 * 8.314 * (-0.693)
delta_S_mix = 11.5 J/K

The entropy of the system rises by about 11.5 J/K when the two gases mix. The number is positive, as the formula guarantees, and it confirms the everyday observation: the gases blend on their own and will never spontaneously separate themselves. To reverse this mix and rebuild the original two pure gases, an external process would have to remove at least this much entropy from the system — and that costs work.

Common mistakes

Thinking mixing releases or absorbs heat for ideal gases. For ideal gases at constant temperature, mixing involves no enthalpy change — the molecules do not interact. The entire driving force is entropic. The blending is spontaneous purely because the mixed state is far more probable, not because energy is lowered by contact.

Applying the formula to identical gases. Mixing a gas with more of the same gas changes nothing measurable — there is no new disorder, because the molecules were already indistinguishable. The formula applies only to genuinely different species. Pretending two samples of the same gas "mix" leads to the classic Gibbs paradox.

Forgetting that the result scales with total moles. delta_S_mix carries an n_total factor. Doubling the amount of every gas doubles the entropy of mixing. The mole fractions set the per-mole part; n_total sets the size.

Assuming all mixtures behave ideally. Real liquids and non-ideal gases have interaction energies, so their mixing entropy departs from the ideal formula and an enthalpy term appears. The ideal expression is the clean baseline; real systems need activity coefficients to correct it.

Reading a positive entropy change as "always favourable" without temperature. Spontaneity is decided by Gibbs free energy, delta_G = delta_H - T*delta_S. For ideal-gas mixing delta_H is zero, so mixing is always spontaneous — but in real systems a positive delta_S can still be outweighed by an unfavourable delta_H.

Try the interactive NovaSolver calculator

The arithmetic for two equal components is quick by hand; seeing how the entropy responds across three components at once is where a tool earns its keep. The Mixing Entropy Simulator — Ideal Gas Three-Component Mixing on NovaSolver computes delta_S_mix in real time from the total moles, the three mole fractions, and the temperature. It also reports the Gibbs free energy of mixing delta_G_mix, the equimolar maximum delta_S_max, and the ratio of your mixture's entropy to that maximum — and visualises the blend as moving particles plus a ternary composition diagram.

Related calculators

Carnot Cycle Simulator — to see how entropy sets the absolute ceiling on the efficiency of any heat engine.
Ideal Gas Simulator (PVT Surface) — explore how pressure, volume, and temperature interlock on the ideal-gas surface.
Ideal Gas Laws Simulator — work through Boyle's, Charles's, and the combined gas laws that underpin the ideal-gas model.

Browse the full set in the thermal and thermodynamics tools hub.

Closing note

The entropy of mixing is one of thermodynamics' most quietly profound results. A short formula — minus n_total times R times the sum of x times ln x — captures why perfume fills a room, why gases never unmix, and why every separation plant on earth has to pay an energy bill. The formula is always positive because mole fractions are always less than one, and that single sign is the second law speaking. Compute it, watch it peak at the equimolar mixture, and you hold a precise measure of nature's preference for the spread-out over the sorted.

Logistic Regression: Turning a Linear Score into a Probability

NovaSolver — Wed, 12 Aug 2026 00:17:02 +0000

A loan application lands on a desk. The reviewer wants a number — not a vague "looks risky," but a probability: this applicant has a 12 percent chance of default, that one a 78 percent chance. The inputs are ordinary measurements like income, debt, and credit history. The output has to be a probability, bounded firmly between 0 and 1. That gap, between a free-ranging weighted sum and a well-behaved probability, is exactly what logistic regression bridges.

This article explains why a straight line cannot output a probability, how the sigmoid function fixes that, and how to read the decision boundary that the model draws.

Why this calculation matters

Logistic regression is the default tool for binary classification across engineering and science: pass or fail, defective or sound, spam or legitimate, disease present or absent. It is the model people reach for first because it is fast to fit, hard to overfit, and — unlike many alternatives — easy to interpret. Each coefficient has a plain meaning in terms of how the odds shift.

It also matters as a foundation. A single artificial neuron with a sigmoid activation is, mathematically, a logistic regression. Understanding this one model well gives you a direct handle on how neural networks make decisions at every node. And because its output is a genuine probability rather than a raw score, logistic regression supports decisions that need a calibrated confidence — where the cost of a false positive and a false negative are not equal, and you want to set the threshold deliberately rather than accept a hard-coded 0.5.

The core method

Start with a linear score, often called the logit. With one predictor x it is just a weighted input plus a bias:

z = b0 + b1*x

That score z can be any real number, from large negative to large positive — which is the problem, because a probability cannot. The fix is the sigmoid, or logistic, function. It takes any real z and squeezes it smoothly into the open interval between 0 and 1:

p = 1 / (1 + e^(-z))

When z is large and positive, e^(-z) shrinks toward zero and p approaches 1. When z is large and negative, e^(-z) blows up and p approaches 0. When z is exactly 0, p is exactly 0.5. The curve is a smooth S: nearly flat at the extremes, steepest in the middle.

The point where p = 0.5 is the decision boundary. It is where z = 0, the dividing surface between the two predicted classes. For the one-variable model, set b0 + b1*x = 0 and solve for x. With more predictors, z = 0 defines a line, a plane, or a hyperplane through the feature space.

The coefficients are not fitted by least squares. They are chosen to maximise the likelihood of the observed labels — equivalently, to minimise the cross-entropy loss — and that optimisation is solved numerically, usually with gradient descent. The sign of each coefficient tells the story: a positive b1 means larger x pushes the prediction toward the positive class.

A worked example

Take a fitted model with intercept b0 = -4 and slope b1 = 2. The task is to predict the probability of the positive class at x = 3.

Step 1 — compute the linear score.

z = b0 + b1*x = -4 + 2*3 = -4 + 6 = 2

Step 2 — pass the score through the sigmoid.

p = 1 / (1 + e^(-z)) = 1 / (1 + e^(-2))

The value of e^(-2) is about 0.135, so:

p = 1 / (1 + 0.135) = 1 / 1.135 = 0.881

The model predicts an 88 percent probability that this case belongs to the positive class. With a standard 0.5 threshold, it is classified positive — and confidently so.

Step 3 — locate the decision boundary. The boundary sits where p = 0.5, which means z = 0:

b0 + b1*x = 0
-4 + 2*x = 0
x = 2

So any input above x = 2 is predicted positive, anything below it negative, and x = 3 sits one full unit into positive territory — consistent with the 88 percent figure. Notice the slope's role: b1 = 2 controls how sharply the probability swings as x crosses the boundary. A larger b1 makes the S-curve steeper and the model more decisive near x = 2.

Common mistakes

Treating the output score as a probability before the sigmoid. The linear score z is not a probability — it is unbounded and can be negative. Only after passing through the sigmoid does the value become a probability between 0 and 1. Skipping that step is a common slip when reading model internals.

Assuming the 0.5 threshold is sacred. Classifying at p = 0.5 is a convention, not a requirement. If a missed positive is far costlier than a false alarm, lower the threshold. Logistic regression hands you a probability precisely so you can choose the cutoff that fits the problem.

Interpreting coefficients as probabilities. A coefficient b1 acts on the log-odds, not on p directly. Because the sigmoid is nonlinear, the same one-unit change in x shifts the probability a lot near the boundary and very little out at the flat tails.

Fitting it like a linear regression. There is no closed-form least-squares solution. The coefficients come from maximising likelihood, solved iteratively. A learning rate that is too large can make the optimisation diverge instead of settle.

Forgetting that perfectly separable data breaks the fit. If a single feature splits the two classes cleanly with no overlap, the maximum-likelihood coefficients run off toward infinity. Regularisation, such as an L2 penalty, keeps them finite and the model stable.

Try the interactive NovaSolver calculator

Reading the sigmoid formula is one thing; watching a decision boundary form is another. The Logistic Regression (2D Binary Classifier) Simulator on NovaSolver fits a sigmoid model to two-dimensional data by gradient descent and shows the linear decision boundary it learns. You can adjust the learning rate, the number of iterations, and the L2 regularization strength, then read off the training accuracy, the final cross-entropy loss, and the fitted weight and bias — a direct, hands-on view of how the optimisation behaves.

Related calculators

Linear Regression Simulator — the continuous-output cousin, where least squares fits a straight line instead of an S-curve.
Regression Analysis & Curve Fitting — for fitting polynomial and other nonlinear trends to numeric data.
Gaussian Naive Bayes Classifier Simulator — a contrasting probabilistic classifier that models each class with its own distribution.

Browse the full set in the math and statistics tools hub.

Closing note

Logistic regression earns its place by doing one thing cleanly: it converts a linear weighted sum into a probability you can act on. The sigmoid is the whole trick — an S-curve that bounds the output, with z = 0 marking the decision boundary. Compute the score, squeeze it through the sigmoid, and read the probability; then choose a threshold that respects the real costs of each kind of error. Master this model and you have not just a reliable classifier but also the basic unit from which neural networks are built.

Linear Regression by Least Squares: Fitting the Best Straight Line to Data

NovaSolver — Tue, 11 Aug 2026 12:17:01 +0000

A lab notebook records the weight a spring carries and how far it stretches. The numbers never fall on a perfect line — the ruler wobbles, the spring warms up, the reading is rounded. Yet there is clearly a trend, and you want a single equation that captures it well enough to predict the stretch at a load you never tested. That is the everyday job of linear regression.

This article explains what "best fit" actually means, derives the least-squares slope and intercept, works a full numerical example by hand, and points out the mistakes that quietly corrupt regression results.

Why this calculation matters

Linear regression is the workhorse of applied data analysis. It calibrates sensors, estimates material constants from test data, projects demand from past sales, and gives the baseline that every more elaborate model is measured against. When an engineer says "the response is roughly linear over this range," a regression line is the object behind that sentence.

It matters because eyeballing a line through a scatter plot is unreliable and unrepeatable. Two people will draw two different lines, and neither can defend the choice. Least squares replaces judgment with a definite rule: out of every possible straight line, pick the one that makes the total squared vertical error as small as possible. That rule produces one answer, the same answer every time, and it comes with diagnostics — the correlation coefficient and the residuals — that tell you whether a straight line was a sensible choice at all.

The core method

A straight-line model has two unknowns, a slope and an intercept:

y = a + b*x

For each data point, the residual is the gap between the observed y and the value the line predicts: residual = y - (a + b*x). Least squares chooses a and b to minimise the sum of those residuals squared. Squaring matters: it makes every error positive so they cannot cancel, and it penalises large misses far more than small ones.

Setting the derivatives of that sum to zero gives two clean formulas. The slope is the cross-product of the deviations divided by the spread in x:

b = sum[(x - x_mean)(y - y_mean)] / sum[(x - x_mean)^2]

The intercept then follows from the fact that the best-fit line always passes through the centroid of the data, the point (x_mean, y_mean):

a = y_mean - b*x_mean

Two quantities help you judge the fit afterward. The correlation coefficient r runs from -1 to +1 and measures how tightly the points hug a line. The coefficient of determination, R squared, is the fraction of the variation in y that the line explains. An R squared of 0.95 means the line accounts for 95 percent of the up-and-down movement in the data; the rest is scatter the model does not capture.

A worked example

Take four points: (1, 2), (2, 5), (3, 7), and (4, 10). The goal is the least-squares line y = a + b*x.

Step 1 — find the means.

x_mean = (1 + 2 + 3 + 4) / 4 = 2.5
y_mean = (2 + 5 + 7 + 10) / 4 = 6.0

Step 2 — form the deviations and the cross-product sum. Subtract each mean, then multiply the paired deviations:

(1-2.5)(2-6) = (-1.5)(-4) = 6.0
(2-2.5)(5-6) = (-0.5)(-1) = 0.5
(3-2.5)(7-6) = ( 0.5)( 1) = 0.5
(4-2.5)(10-6)= ( 1.5)( 4) = 6.0

sum[(x - x_mean)(y - y_mean)] = 6 + 0.5 + 0.5 + 6 = 13

Step 3 — sum the squared x-deviations.

(-1.5)^2 + (-0.5)^2 + (0.5)^2 + (1.5)^2
= 2.25 + 0.25 + 0.25 + 2.25 = 5.0

Step 4 — compute the slope.

b = 13 / 5 = 2.6

Step 5 — compute the intercept.

a = y_mean - b*x_mean = 6.0 - 2.6*2.5 = 6.0 - 6.5 = -0.5

The best-fit line is:

y = -0.5 + 2.6x

Read it back: each unit increase in x raises y by 2.6, and the line crosses the y-axis at -0.5. Check the centroid — at x = 2.5, the line gives -0.5 + 2.6*2.5 = 6.0, exactly y_mean, as it must. The fit is excellent here; the points sit very close to the line.

Common mistakes

Reversing x and y. Regression of y on x is not the same as regression of x on y. Least squares minimises the error in one chosen direction — vertical for y on x. Swap the roles and you get a different slope. Decide which variable you are predicting before you compute anything.

Reading slope as causation. A strong fit shows that two quantities move together, not that one drives the other. Ice cream sales and drowning rates correlate beautifully, and neither causes the other — summer does. Regression quantifies association; causation needs an experiment or a mechanism.

Letting an outlier set the line. Because errors are squared, one stray point far from the trend can pull the whole line toward itself. Always look at the scatter plot and the residuals; a single bad reading deserves investigation, not silent inclusion.

Extrapolating past the data. A line fitted between x = 1 and x = 4 says nothing reliable about x = 40. The linear relationship is an observation within the tested range, not a law that holds forever.

Trusting the slope while ignoring R squared. A line can always be computed, even through a shapeless cloud. A low R squared is the model telling you a straight line is the wrong shape — perhaps the trend is curved, or there is simply no relationship.

Try the interactive NovaSolver calculator

Working one fit by hand fixes the idea; building intuition for how a single point shifts the line is faster with a live tool. The Linear Regression Simulator on NovaSolver lets you click anywhere on the chart to add data points and watch the least-squares line, slope b, intercept a, correlation r, R squared, and RMSE update in real time. You can load sample datasets, toggle the residual segments and the 95 percent band, and drop in an outlier to see exactly how far it drags the fit.

Related calculators

Logistic Regression (2D Binary Classifier) Simulator — when the outcome is a yes/no class rather than a continuous number, the straight line gives way to an S-curve.
Regression Analysis & Curve Fitting — fit polynomial and other nonlinear curves when the residuals show the trend is not a straight line.
Monte Carlo Statistics Simulator — to see how random sampling builds up distributions and how uncertainty behaves.

Browse the full set in the math and statistics tools hub.

Closing note

Least-squares linear regression is small enough to do on paper and powerful enough to anchor most of applied statistics. Find the means, form the cross-products, divide, and you have a slope; the intercept comes free from the centroid. The harder discipline is afterward: look at the residuals, check the correlation, resist the pull of outliers, and never extrapolate beyond the data you actually have. Get the line and read its diagnostics honestly, and a scatter of measurements becomes a model you can defend.

Solar Cell Fill Factor: The Number That Tells You How Good a Cell Really Is

NovaSolver — Tue, 11 Aug 2026 00:17:02 +0000

Two solar cells can share the same open-circuit voltage and the same short-circuit current and still deliver very different amounts of power. One looks crisp; the other is quietly leaking performance. The difference does not show up in the headline numbers — it hides in the shape of the cell's current-voltage curve. The fill factor is the single number that captures that shape, and it is one of the most telling diagnostics in photovoltaics.

This article explains what the fill factor measures, how to calculate it from four points on the I-V curve, and why a value far below 1 is a warning sign about a cell's internal health.

Why this calculation matters

A solar cell's job is to deliver power, and power is voltage times current. But a cell does not produce its peak voltage and peak current at the same time. Short the cell and you get maximum current but zero voltage; open it and you get maximum voltage but zero current. Useful power lives somewhere between those extremes, at the maximum power point.

The fill factor measures how much of the theoretically available rectangle of voltage and current the cell actually captures. That makes it the third pillar of cell performance, alongside open-circuit voltage and short-circuit current. Two of those three are set largely by the semiconductor material and the light. The fill factor, by contrast, is set by the cell's parasitic resistances and recombination — by manufacturing quality. When engineers compare cells, debug a production line, or track degradation in the field, the fill factor is often the number that tells the real story.

The core formula

The fill factor is defined as the ratio of two products:

FF = (V_mp * I_mp) / (V_oc * I_sc)

The numerator is the power at the maximum power point: V_mp and I_mp are the voltage and current where the product V times I is largest. The denominator is the product of open-circuit voltage V_oc and short-circuit current I_sc — a power the cell could never actually deliver, since those two values occur at opposite ends of the curve.

The geometric picture makes it clear. Plot current against voltage. The maximum-power-point operating condition defines a rectangle of area V_mp times I_mp. The values V_oc and I_sc define a larger bounding rectangle of area V_oc times I_sc. The fill factor is the ratio of the inner rectangle to the outer one — literally how much of the available box the cell fills.

FF = (area of MPP rectangle) / (area of Voc-Isc rectangle)

Because the MPP rectangle always fits inside the bounding one, the fill factor is always between 0 and 1. A perfectly square I-V curve — current holding flat right up to V_oc, then dropping vertically — would give FF = 1. Real cells round off at that corner, so the fill factor is always less than 1. Good crystalline-silicon cells reach roughly 0.70 to 0.85. The closer to 1, the squarer the curve and the higher the quality.

Once you have the fill factor, the cell's maximum power follows directly:

P_max = V_mp * I_mp = FF * V_oc * I_sc

This is why the fill factor matters in practice: it converts two easy-to-measure quantities into the power the cell will actually deliver.

A worked example

Take a single solar cell characterized in the laboratory. Its open-circuit voltage is V_oc = 0.60 V and its short-circuit current is I_sc = 8.0 A. Tracing the I-V curve, the maximum power point sits at V_mp = 0.48 V and I_mp = 7.5 A.

Step 1 — the maximum-power-point product.

V_mp * I_mp = 0.48 * 7.5 = 3.6 W

This is the actual power the cell delivers at its best operating point.

Step 2 — the bounding product.

V_oc * I_sc = 0.60 * 8.0 = 4.8 W

This is the idealized rectangle — the power the cell would give if it held full current all the way to full voltage.

Step 3 — the fill factor.

FF = (V_mp * I_mp) / (V_oc * I_sc)
FF = 3.6 / 4.8
FF = 0.75

The fill factor is 0.75. The cell captures 75 percent of the available voltage-current box — a respectable result for a crystalline-silicon cell. The remaining 25 percent is lost to the rounding of the I-V curve, which in turn comes from series resistance, shunt leakage, and recombination inside the device. Notice that V_mp is below V_oc and I_mp is below I_sc; both shortfalls are unavoidable, and the fill factor packages them into one honest number.

Common mistakes

Reading the fill factor as an efficiency. It is not. Efficiency compares electrical output to incident sunlight; the fill factor compares the MPP power to the V_oc-I_sc product and says nothing about how much light arrived. A cell can have a fine fill factor and poor efficiency, or the reverse.

Expecting a fill factor of 1. No real cell reaches it. The corner of the I-V curve always rounds because the diode turns on gradually and parasitic resistances bleed power. Treat values around 0.70 to 0.85 as healthy for silicon and be suspicious of anything claimed much higher.

Ignoring series resistance. A higher series resistance tilts the high-current part of the curve, pushing the maximum power point inward and cutting the fill factor directly. A drifting fill factor in a production batch often points straight at contact or metallization problems.

Forgetting shunt resistance. Leakage paths across the cell — a low shunt resistance — flatten the low-voltage part of the curve and also drag the fill factor down. Series and shunt resistance damage the fill factor from opposite ends of the I-V curve.

Comparing cells at different conditions. The fill factor shifts with temperature and irradiance. Comparing one cell measured hot against another measured cool is not a fair test; line up the conditions before drawing conclusions.

Try the interactive NovaSolver calculator

Seeing the fill factor as a number is one thing; watching the I-V curve change shape as resistances and temperature move is what builds real intuition. The Photovoltaic Cell I-V Characteristics tool on NovaSolver simulates a cell with the single-diode model: set irradiance, cell temperature, ideality factor, series and shunt resistance, and it draws the I-V and P-V curves, marks the maximum power point, and reports maximum power, V_mp, fill factor, and conversion efficiency in real time.

Related calculators

Solar Panel Calculator — scales up from the single cell to a full panel and installation, estimating annual energy, savings, and payback.
Solar Radiation calculator — estimates the irradiance reaching a cell, the input that sets short-circuit current and feeds the fill-factor calculation.
Semiconductor PN Junction tool — explores the diode physics underneath the solar cell, where the I-V curve's shape and its fill factor ultimately originate.

You can browse the rest in the environment and energy tools hub.

Closing note

The fill factor is a small calculation with a large diagnostic reach. It is just the ratio of two areas — the maximum-power rectangle over the V_oc-I_sc rectangle — but that ratio tells you how square a cell's I-V curve is, and squareness is a direct read on internal quality. Voltage and current describe what a cell could do at its extremes; the fill factor describes how well it performs in between. Compute it, watch how series and shunt resistance pull it down, and you have a reliable lens for judging any photovoltaic device.

Wind Turbine Power: Why Output Scales With the Cube of Wind Speed

NovaSolver — Mon, 10 Aug 2026 12:17:02 +0000

Stand beneath a large wind turbine on a breezy day and the blades look almost lazy, sweeping the air maybe fifteen times a minute. It is hard to believe that a single rotor of that size can power hundreds of homes. The secret is not in the speed of the blades but in the volume of air they intercept, and in one unforgiving fact: the power available in wind grows with the cube of its speed.

This article explains where that cube comes from, how to compute the power a turbine actually captures, and why physics places a firm cap on how much of the wind's energy any rotor can ever take.

Why this calculation matters

Wind speed is the single most important variable in any wind project, and the cubic relationship is the reason. A site that averages 8 m/s instead of 6 m/s does not yield 33 percent more energy — it yields closer to twice as much. That sensitivity drives where turbines are sited, how tall the towers are built, and whether a project is financed at all.

The power equation is also the tool for matching a machine to a site. Choose a rotor diameter, estimate the local wind, and you have a first-order figure for annual output and revenue. Get the calculation wrong and the consequences are expensive: an undersized rotor leaves energy in the air, while an oversized one adds structural cost and loads that the site's wind never pays back. Every capacity-factor estimate and every levelized-cost figure traces back to this one relationship.

The core formula

A wind turbine extracts a fraction of the kinetic energy carried by the air passing through its rotor. The captured power is:

P = 0.5 * rho * A * V^3 * Cp

Here rho is the air density in kg/m^3, A is the rotor swept area in m^2, V is the wind speed in m/s, and Cp is the power coefficient — the dimensionless fraction of the wind's power that the rotor actually converts.

The swept area is the disc the blade tips trace out:

A = pi * R^2

where R is the rotor radius, half the rotor diameter.

Each piece has a clear physical reading. The mass of air flowing through the disc each second is proportional to rho times A times V. The kinetic energy per unit mass is proportional to V squared. Multiply the two and the available power carries V cubed — one factor of V for how much air arrives, two more for the energy each parcel carries. That is why wind power is so brutally sensitive to speed: doubling V multiplies the available power by eight.

The power coefficient Cp accounts for the fact that a rotor cannot take all of the wind's energy. If it did, the air would have to stop dead behind the blades, and stopped air cannot make way for the air behind it. The German physicist Albert Betz showed in 1919 that the theoretical maximum is:

Cp_max = 16 / 27 = 0.593

This is the Betz limit. No turbine, however cleverly designed, can convert more than about 59.3 percent of the wind's kinetic power. Good modern machines reach Cp values in the range of 0.40 to 0.50 — close to the limit, but never at it.

A worked example

Take a turbine with a rotor diameter of 50 m. The wind blows steadily at V = 10 m/s, the air density is rho = 1.2 kg/m^3, and the rotor achieves a power coefficient of Cp = 0.40.

Step 1 — swept area.

The rotor radius is R = 25 m, so:

A = pi * (25)^2
A = pi * 625
A = 1963 m^2

The blades sweep nearly two thousand square meters of air — about a third of a soccer pitch standing on end.

Step 2 — captured power.

P = 0.5 * rho * A * V^3 * Cp
P = 0.5 * 1.2 * 1963 * (10)^3 * 0.40
P = 0.5 * 1.2 * 1963 * 1000 * 0.40
P = 471,120 W

So this turbine delivers about 471 kW in a 10 m/s wind.

It is worth pausing on what the cube does here. If the wind picked up to 12 m/s, the V^3 term would rise from 1000 to 1728 — a 73 percent jump in available power from a 20 percent rise in speed. And if this rotor somehow reached the Betz limit of Cp = 0.593 instead of 0.40, the output would climb to roughly 699 kW. The 471 kW figure is real and good; the gap to 699 kW is the slice of the wind that physics simply will not let any rotor have.

Common mistakes

Treating power as linear in wind speed. The most expensive error in wind engineering. Power follows V cubed, so small changes in average wind speed produce large changes in energy yield. A site survey that is off by 1 m/s can swing a project's economics dramatically.

Believing a turbine can capture all the wind's energy. It cannot. The Betz limit caps Cp at 0.593, and real rotors fall short of even that. Any claim of efficiency above 59.3 percent of the wind's kinetic power is physically impossible.

Using a fixed air density. rho depends on altitude, temperature, and humidity. A turbine on a cold lowland site sees denser air — and more power — than the same machine on a warm mountain ridge. The common 1.225 kg/m^3 sea-level value can be noticeably off in either direction.

Forgetting that output is capped above the rated wind speed. The cubic curve does not run forever. Above the rated speed a turbine deliberately sheds power, by pitching its blades, to protect the drivetrain. Real machines also stop entirely at a cut-out speed in storms.

Confusing instantaneous power with annual energy. A single power figure assumes one steady wind speed. Real sites have a distribution of speeds, often modeled with a Weibull curve. Annual energy comes from integrating the power curve over that distribution, not from multiplying one number by 8760 hours.

Try the interactive NovaSolver calculator

Working one wind speed by hand shows the principle, but a real site is a spread of speeds across a whole year. The Wind Turbine Power Simulator on NovaSolver lets you set rotor diameter, power coefficient Cp, rated wind speed, mean wind speed, and the Weibull shape factor, then returns rated power, annual energy production, capacity factor, and how close your Cp sits to the 59.3 percent Betz limit — so you can see the cubic law and the Weibull distribution working together.

Related calculators

Wind Turbine calculator — a broader look at turbine behavior for quick checks on rotor sizing and output.
Wind Turbine Design tool — steps into blade and rotor design choices once the headline power figure is settled.
Solar Panel Calculator — the companion for hybrid systems, where solar generation often fills the calm hours when the wind drops.

You can browse the rest in the environment and energy tools hub.

Closing note

Wind turbine power comes down to one equation and one limit. The equation, P = 0.5 rho A V^3 Cp, says output rises with swept area and, far more steeply, with the cube of wind speed. The limit, Betz's 0.593, says no rotor can ever take more than about 59.3 percent of the wind's energy. Hold both in mind and the behavior of any turbine becomes predictable: chase the windy sites, respect the cube, and treat Cp as something to approach but never reach.

Solar Panel Power Output: How to Estimate What a Panel Actually Delivers

NovaSolver — Mon, 10 Aug 2026 00:17:02 +0000

A solar panel rated at 400 watts almost never delivers 400 watts. Buy one, mount it on a roof, and the real number depends on how bright the sun is at that moment, how big the panel is, and how efficiently the cells turn light into current. On a clear noon it might hit its rating; an hour before sunset it might give a tenth of that. The rating is a laboratory promise, not a field guarantee.

This article works through the basic relationship between sunlight and electrical output, shows how to turn instantaneous power into a daily energy figure, and points out where back-of-the-envelope estimates quietly go wrong.

Why this calculation matters

Sizing a solar installation is fundamentally an energy-balance problem. You have a load — a house, a pump, a remote sensor — that needs a certain number of kilowatt-hours per day. You have a site that receives a certain amount of sunlight. The panel calculation is the bridge between the two, and getting it wrong propagates through everything downstream.

Undersize the array and the batteries drain on cloudy afternoons. Oversize it and you have paid for hardware that spends much of its life clipped or curtailed. The same arithmetic decides how many panels fit a roof, how big an inverter needs to be, and how many years it takes for the system to pay for itself. None of those answers are reliable until you can estimate output from first principles rather than trusting a sticker.

The core formula

The instantaneous electrical power a panel produces is the product of three quantities:

P = G * A * eta

Here G is the solar irradiance in watts per square meter — the power density of sunlight striking the panel. A is the panel's area in square meters. And eta is the conversion efficiency, the dimensionless fraction of incident light energy that leaves the panel as electricity.

The logic is straightforward. The term G times A is the optical power landing on the panel. Multiply by eta and you have the electrical power coming out. A standard reference condition uses G = 1000 W/m^2, often called "one sun" or "peak sun," which roughly matches bright midday sunlight at the surface.

To get energy rather than power, multiply by time. Because irradiance changes continuously through the day, engineers collapse the whole solar day into an equivalent number of full-strength hours:

E = P_peak * H

H is the number of peak-sun-hours — the count of hours of 1000 W/m^2 sunlight that would deliver the same total energy as the actual, varying day. A location with H = 5 receives, over a full day, the same energy as five hours of perfect noon sun. This single number folds latitude, season, and weather into one figure you can multiply.

Two things are worth holding onto. Output scales linearly with all three of G, A, and eta, so a 10 percent gain in efficiency is a 10 percent gain in power. And the peak-sun-hour trick only works because energy is the integral of power over time — H is that integral repackaged as an equivalent duration.

A worked example

Consider a single photovoltaic panel with an area of A = 2 m^2 and a conversion efficiency of eta = 0.20, a realistic value for a good crystalline-silicon module. The site receives bright sun, so take G = 1000 W/m^2.

Step 1 — instantaneous power.

P = G * A * eta
P = 1000 * 2 * 0.20
P = 400 W

Under one full sun, this panel delivers 400 watts. That is the rated, best-case number — the figure on the datasheet.

Step 2 — daily energy.

Assume the location offers an effective 5 peak-sun-hours over the day.

E = P * H
E = 400 * 5
E = 2000 Wh = 2.0 kWh

So this one panel generates about 2.0 kWh on a representative day. To make that concrete, that is roughly enough to run a typical refrigerator for a full day, or to fully charge several laptops with energy to spare.

The two results sit at different points on the same curve. The 400 W is the peak the panel can hit; the 2.0 kWh is what an entire day of real, rising-and-falling sunlight adds up to. Confusing the two is the most common mistake in solar estimation.

Common mistakes

Treating the panel rating as the average output. A 400 W rating is measured at 1000 W/m^2 and 25 C. Mornings, evenings, clouds, and high cell temperatures all pull the real output below it. Daily average power is often a small fraction of the nameplate figure.

Multiplying power by 24 hours. The sun is not up all day, and it is not at full strength when it is. Use peak-sun-hours, not clock hours. A site with H = 5 does not produce energy as if the panel ran flat-out for 24 hours, or even for the 12 hours of daylight.

Ignoring temperature. Silicon cells lose efficiency as they heat up — often a fraction of a percent per degree above 25 C. A panel baking on a dark roof in summer can run well above ambient, so its hot-weather output trails its rated value.

Forgetting system losses. The formula gives the panel's DC output. Wiring resistance, inverter conversion, dust, shading, and module mismatch all skim energy off before it reaches the load. Real systems often deliver something like 75 to 85 percent of the ideal panel figure.

Using a single irradiance for the whole year. Peak-sun-hours vary strongly with season and latitude. A winter H can be less than half the summer value, so a system sized to a yearly average may fall short in the months you need it most.

Try the interactive NovaSolver calculator

Running the arithmetic by hand is fine for one panel on one day, but real projects involve many modules and a full year of seasonal variation. The Solar Panel Calculator on NovaSolver takes panel power, number of panels, tilt angle, latitude, system efficiency, electricity price, and installation cost, and returns annual energy in kWh, monthly average generation, annual savings, payback period, CO2 saved, and total system size. It lets you move from a single-panel estimate to a sized, costed installation in a few slider movements.

Related calculators

Photovoltaic Cell I-V Characteristics — drops one level down to the cell itself, showing how irradiance and temperature reshape the current-voltage curve and set the efficiency you plug into the panel formula.
Solar Radiation calculator — helps you estimate the irradiance G and the peak-sun-hours for a given location and orientation, the inputs the panel calculation depends on most.
Wind Turbine Power Simulator — the natural companion for a hybrid renewable system, where wind often produces when the sun does not.

You can explore the rest in the environment and energy tools hub.

Closing note

Solar estimation rests on one compact idea: electrical power is irradiance times area times efficiency, and daily energy is that power multiplied by an equivalent number of full-sun hours. The formula is simple enough to do on paper, which is exactly why it is worth doing — it builds the intuition that a 400 W panel is a peak figure, that a kilowatt-hour is a peak figure stretched over time, and that real systems sit somewhere below both. Start with the clean calculation, then subtract honestly for temperature, weather, and system losses, and your energy budget will hold up.

Optical Fiber Numerical Aperture: How Much Light a Fiber Can Catch

NovaSolver — Sun, 09 Aug 2026 12:17:02 +0000

A single strand of glass thinner than a human hair carries most of the world's internet traffic across oceans. The light that does the carrying never escapes the fiber's core — it bounces along the inside, trapped, for hundreds of kilometres. But before any of that can happen, the light has to get in. Aim a beam at the end face of a fiber from too steep an angle and it simply leaks out the side a few millimetres later, instead of travelling the length of the cable.

This article explains how a fiber traps light, what the numerical aperture measures, how to calculate it from the core and cladding refractive indices, and how to turn that number into the acceptance angle that decides whether your light couples in or is lost.

Why this calculation matters

Numerical aperture, almost always written NA, is the single number that describes how good a fiber is at collecting light. It sets the size of the cone within which incoming rays will be guided, so it directly governs coupling efficiency — how much of a source's output actually makes it into the core. A laser, an LED, or another fiber must deliver its light inside that cone, or the energy is wasted.

NA also shapes the trade-offs a system designer lives with. A large numerical aperture catches more light from a cheap, broad source like an LED, which is convenient. But the same large NA lets in rays at many different angles, and rays at different angles take paths of different length down the fiber, which spreads a pulse out in time. That spreading, called modal dispersion, limits how fast you can send data. NA therefore sits right at the intersection of how much light you can launch and how much bandwidth you get to keep — which is why it appears on essentially every fiber datasheet.

The core method

Light stays inside a fiber by total internal reflection. The fiber has a central core of refractive index n1 surrounded by a cladding of slightly lower index n2. When a ray inside the core strikes the core-cladding boundary at a shallow enough angle, it reflects perfectly back into the core instead of refracting out. Repeat that bounce thousands of times and the light is guided.

For total internal reflection to work, the core index must exceed the cladding index, n1 greater than n2. The numerical aperture captures how large that index difference is, and therefore how wide the guided cone can be:

NA = sqrt(n1^2 - n2^2)

where n1 is the core refractive index and n2 the cladding refractive index. A bigger index contrast gives a bigger NA and a wider acceptance cone.

The numerical aperture connects directly to a physical angle. The maximum acceptance angle — the half-angle of the cone of light, measured from the fiber axis in air, that the fiber will accept and guide — follows from:

theta_max = arcsin(NA)

Any ray entering the end face within theta_max of the axis will strike the core-cladding interface shallowly enough to be totally internally reflected and carried down the fiber. A ray entering outside that cone hits the interface too steeply, refracts into the cladding, and is lost.

A worked example

Take a typical single-mode-style glass fiber with a core refractive index n1 = 1.48 and a cladding refractive index n2 = 1.46. Find the numerical aperture and the maximum acceptance angle.

Step 1 — square each refractive index.

n1^2 = 1.48^2 = 2.1904
n2^2 = 1.46^2 = 2.1316

Step 2 — take the difference and its square root.

NA = sqrt(2.1904 - 2.1316) = sqrt(0.0588) = 0.2425

So the numerical aperture is about 0.24 — a modest value, typical of telecom-grade glass fiber where the core and cladding indices are deliberately kept close together.

Step 3 — convert NA into the acceptance angle.

theta_max = arcsin(NA) = arcsin(0.2425) = 14.0 degrees

The fiber accepts light within a cone of about 14 degrees half-angle. Any beam entering the end face inside that cone is guided down the fiber; anything steeper leaks away into the cladding. That fairly narrow cone is the price of low modal dispersion — and it is why coupling light efficiently into a fiber takes careful alignment rather than a casual aim.

Common mistakes

Subtracting the indices instead of their squares. The formula is the square root of (n1 squared minus n2 squared), not the square root of (n1 minus n2). The difference of squares is what the geometry of total internal reflection actually produces; skipping the squaring gives a wrong, usually larger, NA.

Confusing acceptance angle with the full cone angle. theta_max is the half-angle, measured from the fiber axis. The full angular width of the accepted cone is twice that. Quoting one when you mean the other doubles or halves your aperture.

Assuming the launch medium is always air. The simple arcsin(NA) relation gives the acceptance half-angle in air, where the index is essentially 1. If light enters the fiber through index-matching gel or another medium, that medium's index changes the external angle.

Treating a high NA as automatically better. A larger numerical aperture gathers more light, which helps with LEDs and other broad sources. But it also admits more modes and more modal dispersion, narrowing the usable bandwidth. The best NA depends on whether you are optimizing for coupling or for data rate.

Ignoring the index difference's sensitivity. Because NA depends on a small difference between two nearly equal numbers, a tiny error in n1 or n2 shifts the result noticeably. Use index values precise enough that the subtraction stays meaningful.

Try the interactive NovaSolver calculator

Computing one numerical aperture by hand is quick, but seeing how NA, acceptance angle, dispersion and reach respond to real fiber and system choices is where intuition is built. The Optical Fiber Transmission tool on NovaSolver lets you set fiber type, attenuation, length, launch power, splice and connector counts, receiver sensitivity and dispersion, then estimates the received power, the power margin, the link bandwidth and the maximum reach — so you can watch how design decisions add up across a whole fiber link.

Related calculators

Fiber Optics Calculator — for the core guiding physics of refractive index, total internal reflection and acceptance cone.
Optical Fiber Communication — to extend the link analysis toward data rate, loss budget and system performance.
Snell's Law Calculator — for the refraction and critical-angle physics that total internal reflection is built on.

You can browse the rest in the optics tools hub.

Closing note

Numerical aperture compresses a fiber's light-gathering ability into one number drawn from just two refractive indices. The square root of the difference of their squares gives the NA; the arcsine of the NA gives the acceptance cone. Those two short steps tell you whether a source will couple efficiently into a fiber and hint at the dispersion you will pay for it. Square before you subtract, keep the half-angle and full-angle straight, and remember that a bigger aperture is a trade, not a free upgrade — and the physics of guided light becomes a calculation you can trust.

Malus's Law: How a Polarizer Decides How Much Light Gets Through

NovaSolver — Sun, 09 Aug 2026 00:17:02 +0000

Hold two pairs of polarized sunglasses up to a window and look through both at once. Rotate one pair slowly. At some point the world goes nearly black, even though each lens on its own is perfectly transparent. Nothing absorbed the light suddenly — you simply turned one filter until its axis crossed the other's. That dimming is not a gimmick. It is a precise, predictable physical effect, and the rule that describes it is one of the cleanest equations in optics.

This article explains what polarized light is, how a polarizer filters it, how to calculate the transmitted intensity with Malus's law, and why a third polarizer can make blocked light reappear.

Why this calculation matters

Polarizers are everywhere once you start looking. Every liquid-crystal display sandwiches its pixels between two of them; the screen brightness you see is Malus's law playing out millions of times per frame. Photographers rotate a polarizing filter to cut glare off water and glass. Stress engineers use crossed polarizers to make load paths glow inside transparent plastic models. Optical instruments, laser systems, and 3D cinema all lean on controlling the polarization state of light.

In each case the practical question is the same: given the orientation of a polarizer, how much light comes out the other side? Guessing is not good enough. A display designer needs the contrast ratio, an optical engineer needs the power budget, and a measurement setup needs to know how much signal survives each element. Malus's law turns that question into one short calculation, and understanding it also explains a result that surprises almost everyone the first time they see it.

The core formula

Light is a transverse electromagnetic wave: its electric field oscillates perpendicular to the direction of travel. In unpolarized light — sunlight, a bare bulb — that oscillation points in every direction at random. A polarizer is a filter that transmits only the field component aligned with its own axis and removes the rest.

When unpolarized light hits an ideal polarizer, exactly half the intensity gets through, and what emerges is fully polarized along the polarizer's axis. The interesting calculation begins with the second polarizer, because now the incoming light already has a definite polarization direction.

Let theta be the angle between the polarization direction of the incoming light and the axis of the polarizer. Malus's law gives the transmitted intensity:

I = I0 * cos^2(theta)

Here I0 is the intensity of the polarized light arriving at the polarizer and I is the intensity leaving it. The reason for the cosine squared is worth holding onto. The polarizer passes the field component along its axis, and that component scales as cos(theta). Intensity is proportional to the square of the field amplitude, so the transmitted intensity scales as cos squared of the angle.

Two limits anchor the rule. When theta = 0, the axes are aligned, cos(0) = 1, and all the light passes. When theta = 90 degrees, the axes are crossed, cos(90) = 0, and nothing passes — the dark window from the sunglasses experiment.

A worked example

Suppose fully polarized light of intensity I0 strikes a polarizer whose axis is rotated 30 degrees away from the light's polarization direction. How much gets through?

Step 1 — take the cosine of the angle.

cos(30 degrees) = 0.866

Step 2 — square it.

cos^2(30 degrees) = (0.866)^2 = 0.75

Step 3 — apply Malus's law.

I = I0 * cos^2(30 degrees) = I0 * 0.75 = 0.75 * I0

So 75% of the light is transmitted. A 30-degree misalignment costs only a quarter of the intensity — the cosine-squared curve is fairly flat near zero angle, so small misalignments are forgiving.

Now push the angle to 90 degrees. There cos(90) = 0, so I = I0 * 0 = 0: with crossed polarizers the transmitted intensity falls to zero and the field is completely blocked. Between those two points the curve falls steeply — at 45 degrees, cos squared is exactly 0.5, so half the light survives. That midpoint is why a polarizer set at 45 degrees is such a useful "half-power" element.

Common mistakes

Using cosine instead of cosine squared. A frequent slip. The field amplitude scales as cos(theta), but the quantity you measure — intensity or power — scales as cos squared. Forgetting the square overstates the transmitted light, badly at large angles.

Working in degrees and radians inconsistently. Most calculators and programming languages expect radians. Feeding 30 into a function that wants radians returns nonsense. Convert, or set the calculator to degree mode and confirm it.

Applying Malus's law to the first polarizer. The cos-squared rule assumes the incoming light is already polarized. Unpolarized light hitting the first polarizer simply loses half its intensity regardless of orientation; Malus's law governs every polarizer after that.

Expecting an ideal block. Real polarizers leak. A small fraction of light passes even when fully crossed, set by the extinction ratio of the device. For high-contrast work that residual transmission matters.

Misreading the three-polarizer result. Put a polarizer between two crossed polarizers and light reappears, which feels like it should be impossible. It is not a violation of anything. Each polarizer re-projects the polarization onto a new axis, so two 45-degree steps deliver 0.5 times 0.5 = 0.25 of the light, where a single 90-degree step delivers zero.

Try the interactive NovaSolver calculator

The three-polarizer paradox is far easier to believe once you can watch the numbers move. The Malus Law Simulator — Three-Polarizer Paradox on NovaSolver lets you set the input intensity and the angles of three stacked polarizers, then computes the intensity after each plate, the total transmittance, and the contribution of the middle polarizer — so you can dial the centre plate in and out and see crossed-polarizer light switch back on in real time.

Related calculators

Polarization of Light — for the broader picture of linear, circular and elliptical polarization states.
Optical Fiber Transmission — to follow polarized light into a guided medium and estimate received power and reach.
Diffraction Grating Calculator — for another wave-optics workhorse, splitting light by wavelength into its spectral orders.

You can browse the rest in the optics tools hub.

Closing note

Malus's law is small enough to memorize and deep enough to keep teaching you. Intensity through a polarizer equals the incoming intensity times the cosine squared of the angle between the light's polarization and the polarizer's axis — that is the whole rule. It sets the brightness of every LCD pixel, the cut of every polarizing filter, and the surprising glow of the three-polarizer experiment. Remember the square, mind your angle units, and the behaviour of polarized light stops being magic and becomes arithmetic.

Surface Tension and Capillary Action: Why Liquids Climb Thin Tubes

NovaSolver — Sat, 08 Aug 2026 12:17:03 +0000

Dip a strip of paper towel into a puddle and watch the wet line creep upward against gravity. Stand a thin glass tube in a dish of water and the water inside rises noticeably above the surface outside. Nobody pushed it. No pump, no pressure source — just the liquid itself, climbing on its own. The same effect carries sap up the stem of a plant, draws ink from a fountain pen onto paper, and lets a tree wick moisture from soil through roots finer than a hair.

This article explains what surface tension is, why it produces capillary rise, how to calculate the rise height for a given liquid and tube, and where the calculation quietly goes wrong.

Why this calculation matters

Capillary action is not a laboratory curiosity. It governs how fuel moves through a wick, how moisture migrates through concrete and soil, how solder flows into a joint, and how a heat pipe returns its working fluid to the evaporator without any moving parts. Microfluidic chips route droplets through channels narrower than a human hair using nothing but surface forces. Get the surface-tension balance wrong and the device simply does not move liquid where it needs to go.

It also matters in the other direction. In a packed column, a soil sample, or a porous filter, capillary forces can trap liquid you are trying to drain, or pull liquid into places you want to keep dry. Knowing the rise height tells you how tall a wick must be, how fine a pore can be before water bridges it, and whether a coating will self-level or bead up. The calculation is short, but it sits underneath a lot of real engineering decisions.

The core formula

Surface tension, written gamma, is the energy cost of creating new liquid surface, measured in newtons per metre. Molecules inside a liquid are pulled equally in all directions; molecules at the surface are pulled inward, so the surface behaves like a stretched elastic skin that resists being enlarged.

When a liquid meets a solid wall, a second quantity enters: the contact angle theta, the angle the liquid surface makes with the wall. A small contact angle means the liquid wets the solid and is pulled along it; a large angle means the liquid is repelled. Water on clean glass has a contact angle near zero; mercury on glass has an angle well above 90 degrees, which is why mercury is pushed down a tube rather than up.

In a narrow circular tube, the wetted rim pulls the liquid column upward until that pull is balanced by the weight of the raised liquid. Setting the two equal gives the capillary rise height:

h = 2 * gamma * cos(theta) / (rho * g * r)

Here h is the rise height, gamma the surface tension, theta the contact angle, rho the liquid density, g the acceleration of gravity, and r the inner radius of the tube. The result, often called Jurin's law, carries one striking message: rise height is inversely proportional to tube radius. Halve the radius and the liquid climbs twice as high.

A companion quantity is the Laplace pressure, the pressure jump across the curved liquid surface:

delta_p = 2 * gamma / R_curvature

where R_curvature is the radius of the curved meniscus. The tighter the curve, the larger the pressure difference — the same physics that makes a small bubble harder to inflate than a large one.

A worked example

Take water rising in a fine glass capillary tube. Use surface tension gamma = 0.0728 N/m, and assume the water wets the clean glass so the contact angle theta is close to zero, which makes cos(theta) = 1. The tube has an inner radius r = 0.5 mm = 5e-4 m. Water density is rho = 1000 kg/m^3, and g = 9.81 m/s^2.

Step 1 — assemble the numerator.

2 * gamma * cos(theta) = 2 * 0.0728 * 1 = 0.1456

Step 2 — assemble the denominator.

rho * g * r = 1000 * 9.81 * 5e-4 = 4.905

Step 3 — divide to get the rise height.

h = 0.1456 / 4.905 = 0.0297 m

So the water climbs about 30 mm — three centimetres — above the level in the surrounding dish, with no external help at all. That is a substantial height for a tube only one millimetre across.

The radius dependence is worth feeling directly. Shrink the tube to r = 0.1 mm and the rise jumps to roughly 150 mm. Open it up to r = 5 mm and the rise drops to about 3 mm. The narrower the tube, the higher the liquid rises — which is exactly why capillary effects dominate in fine pores and fabrics and become almost invisible in wide pipes.

Common mistakes

Using diameter where the formula wants radius. The equation takes the tube radius r, not the diameter. Slipping in the diameter halves the predicted rise. When in doubt, check the units and sanity-check against the expected order of magnitude.

Forgetting the contact angle entirely. The cos(theta) term is not optional. For water on clean glass it is close to one and easy to overlook, but for a partially wetting surface theta might be 60 degrees, cutting the rise by half. For a non-wetting pair the cosine goes negative and the liquid is depressed, not raised.

Assuming surface tension is a fixed constant. Gamma drops as temperature rises, and even a trace of surfactant or contamination can lower it sharply. A value measured for pure water at 20 C will not describe soapy water or hot water.

Applying Jurin's law to wide tubes. The simple formula assumes the meniscus is a near-perfect spherical cap, which holds only when the tube is narrow compared with the capillary length. In a wide tube the meniscus flattens and the equation overpredicts the rise.

Mixing unit systems for surface tension. Surface tension is often quoted in mN/m or dyn/cm. Water's 72.8 mN/m is 0.0728 N/m. Dropping the conversion throws the answer off by a factor of a thousand.

Try the interactive NovaSolver calculator

Working one case by hand is instructive, but a feel for how rise height trades off against tube size and wetting comes faster when the numbers update live. The Surface Tension & Capillary Rise Simulator on NovaSolver lets you pick a liquid preset or enter custom surface tension, density, contact angle and tube radius, then returns the capillary rise height, the Laplace pressure difference and the Bond number, alongside a diagram that redraws the meniscus and droplet contact-angle shape as you adjust the inputs.

Related calculators

Archimedes' Buoyancy Calculator — for the other classic surface-and-fluid balance, the upward force on a submerged or floating body.
Hydrostatic Pressure Calculator — to see how pressure builds with depth, the weight term that capillary rise has to overcome.
Fluid Pressure Calculator — for converting between pressure, force and area when sizing wicks, seals and small passages.

You can browse the rest in the fluid dynamics tools hub.

Closing note

Surface tension and capillary action turn a liquid's own skin into a pump. The governing relation is compact — rise height equals twice the surface tension times the cosine of the contact angle, divided by density, gravity and tube radius — but it explains a remarkable range of behaviour, from sap in a stem to solder in a joint. Keep three things straight: use the radius, never forget the contact angle, and watch your units on gamma. Do that, and the climb of a liquid up a thin tube becomes a number you can predict instead of a trick you merely admire.

Hydrostatic Pressure: Why Depth Alone Decides the Squeeze

NovaSolver — Sat, 08 Aug 2026 00:17:02 +0000

Swim down to the bottom of a deep pool and your ears begin to ache within a few metres. Go deeper and the ache sharpens. The water has not changed; what changes is the column of water stacked above you, each layer pressing down on the one below. By the floor of the Mariana Trench, eleven kilometres down, that column squeezes with the force of more than a thousand atmospheres — enough to crush an unprotected submarine like a drinks can.

This article explains where that pressure comes from, how to calculate it at any depth, and why a result that surprises many people is actually true: the pressure depends only on depth, not on whether you are in a narrow pipe or a vast lake.

Why this calculation matters

Hydrostatic pressure sets the loads on a huge range of structures and equipment. Dam walls are built thick at the base and thin at the crest precisely because the water pressure they hold back grows with depth. The walls of a storage tank, the hull of a submarine, the casing of a deep-sea sensor, and the glass of an aquarium are all sized against the pressure the fluid exerts on them.

It also drives everyday systems that are easy to overlook. A water tower delivers pressure to a whole neighbourhood by sitting high above it. A medical IV drip relies on the bag hanging above the patient. A manometer reads pressure as a height of liquid. In every one of these cases, getting the depth-to-pressure relationship right is the difference between a design that holds and one that leaks, bursts, or fails to deliver. Because the relationship is a simple linear one, the calculation is quick — but only if you apply it correctly.

The core formula

In a fluid at rest, the gauge pressure increases linearly with depth:

P = rho * g * h

Here rho is the fluid density in kilograms per cubic metre, g is gravitational acceleration (about 9.81 m/s^2), and h is the depth below the free surface in metres. The pressure comes out in pascals.

This formula gives the gauge pressure — the pressure relative to the atmosphere already pressing on the surface. To get the absolute pressure, add the pressure at the surface:

P_absolute = P_0 + rho * g * h

where P_0 is the pressure at the free surface, often atmospheric pressure, about 101.3 kPa.

The reason the law is so simple is worth understanding. Imagine an imaginary column of fluid reaching from the surface down to depth h. That column has weight, and the only thing holding it up is the pressure of the fluid beneath it. Set the upward pressure force equal to the weight of the column, divide by the column's cross-sectional area, and the area cancels out completely. What remains is rho times g times h.

That cancellation is the key insight: pressure at a given depth depends only on the density, gravity, and depth — never on the width or shape of the container. A thimble of water and a reservoir, at the same depth, are at the same pressure. This is sometimes called the hydrostatic paradox, though there is nothing paradoxical about it once the column argument is clear. The same principle means pressure at one depth acts equally in all directions, which is why it pushes outward on a tank wall just as hard as it pushes down on the floor.

A worked example

Take a point 10 metres below the surface of a body of fresh water. The density of water is rho = 1000 kg/m^3, and g = 9.81 m/s^2. Find the gauge pressure at that depth.

Step 1 — identify the variables. The depth is h = 10 m, the fluid density is rho = 1000 kg/m^3, and gravity is g = 9.81 m/s^2.

Step 2 — apply the hydrostatic equation.

P = rho * g * h
P = 1000 * 9.81 * 10
P = 98,100 Pa

Step 3 — interpret the result. The gauge pressure at 10 m depth is 98,100 Pa, or about 98 kPa. That is remarkably close to one standard atmosphere (101.3 kPa). This gives a useful rule of thumb: every 10 metres of water depth adds roughly one atmosphere of pressure. A diver at 10 m feels about twice the absolute pressure they felt at the surface — atmospheric plus one more atmosphere from the water.

Notice what did not appear in the calculation. There is no term for the width of the water body, the shape of the container, or the total volume of water. A diver 10 m down in a narrow flooded shaft feels exactly the same 98 kPa as a diver 10 m down in the open ocean. Only depth, density, and gravity matter.

Common mistakes

Thinking a wider container means more pressure. It does not. Pressure at a given depth is set by depth alone. A tall, thin tube of water and a broad shallow tank produce the same pressure at the same depth. The cross-sectional area cancels out of the derivation.

Mixing up gauge and absolute pressure. The formula rho times g times h gives gauge pressure, measured relative to the atmosphere. If you need absolute pressure — for gas-law calculations or for the true load on a sealed vessel — add atmospheric pressure. Reporting one when the other is wanted can throw a result off by a full atmosphere.

Measuring depth from the wrong reference. The h in the equation is the depth below the free surface of the fluid, not the height of the tank or the distance from the ground. For a partly filled tank, measure down from the liquid level.

Using the wrong fluid density. Seawater (about 1025 kg/m^3) gives a higher pressure than fresh water at the same depth, and mercury (13,600 kg/m^3) is dramatically higher still. Always match the density to the actual fluid.

Forgetting that pressure acts in all directions. Hydrostatic pressure is not just a downward push. At any point it presses equally in every direction, which is why it loads the vertical walls of a tank, not only the floor.

Try the interactive NovaSolver calculator

Computing one depth is easy; building a feel for how pressure climbs through a water column is easier to do visually. The Hydrostatic Pressure Simulator on NovaSolver lets you set the depth and pick a fluid — water, seawater, mercury, or air — and it returns the absolute pressure, the gauge pressure, the pressure expressed in atmospheres, and the force on a one-square-metre surface, alongside a pressure profile you can scan from the surface to the seabed.

Related calculators

Fluid Pressure & Buoyancy Simulator — connects depth-driven pressure to the buoyant force it produces on a submerged body.
Archimedes Buoyancy Simulator — for deciding whether an object floats or sinks once you know the pressure field around it.
Surface Tension & Capillary Rise Simulator — for the small-scale pressure effects that take over when a fluid meets a narrow tube.

You can browse the full set in the fluid mechanics tools hub.

Closing note

Hydrostatic pressure is one of the cleanest results in fluid mechanics: pressure equals density times gravity times depth, and nothing else. The container's shape drops out, the volume drops out, and what remains is a straight line through depth. Three points carry the practical weight: measure depth from the free surface, keep gauge and absolute pressure clearly separate, and match the density to the real fluid. With those in hand, the load on a dam, a tank, or a diver's eardrum becomes a single multiplication.

Archimedes' Principle: Why Things Float, and How to Calculate It

NovaSolver — Fri, 07 Aug 2026 12:17:02 +0000

A steel ship weighs tens of thousands of tonnes, yet it floats. A solid steel bolt of the same metal sinks instantly in the same water. The metal is identical, so the difference cannot be the material. It is the shape — specifically, how much water each object pushes out of the way. The ship is mostly hollow and shoves aside a huge volume of water; the bolt displaces only its own small volume. That single observation is Archimedes' principle, and it governs everything from submarines to hot-air balloons.

This article explains where the buoyant force comes from, how to calculate it, and how to decide whether an object will float, sink, or hover. The arithmetic is simple; the place people slip is in deciding which volume to use.

Why this calculation matters

Buoyancy is the deciding force in any system where a body sits in a fluid. Naval architects use it to set how deep a hull rides and how much cargo it can carry. Submarine designers tune it deliberately, flooding and emptying ballast tanks to dive and surface. It explains why a hot-air balloon climbs, why a hydrometer measures the density of a liquid, and why a diver carries weights to stop bobbing back to the surface.

Get the buoyancy wrong and the consequences are not subtle. Overload a vessel and it rides too low, loses freeboard, and is at risk in rough water. Misjudge the buoyancy of a subsea structure during installation and it can refuse to sink, or sink too fast. Because the buoyant force depends only on displaced volume and fluid density, a quick calculation tells you immediately whether a design is workable before any metal is cut.

The core formula

Archimedes' principle states that the upward buoyant force on a body in a fluid equals the weight of the fluid that the body displaces:

F_b = rho_fluid * g * V_displaced

Here rho_fluid is the density of the fluid in kilograms per cubic metre, g is gravitational acceleration (about 9.81 m/s^2), and V_displaced is the volume of fluid pushed aside, in cubic metres. The force comes out in newtons.

The physical origin is pressure. Fluid pressure increases with depth, so the bottom of a submerged object sits in higher-pressure fluid than the top. The pressure pushing up on the underside therefore exceeds the pressure pushing down on the topside, and the net result is an upward force. Working through the pressure integral over the whole surface gives exactly rho times g times the displaced volume — the weight of the missing fluid.

To predict the object's fate, compare the buoyant force with the object's own weight, W = rho_object times g times V_object:

W < F_b   ->  the object rises and floats
W = F_b   ->  the object is neutrally buoyant and hovers
W > F_b   ->  the object sinks

For a fully submerged body, V_displaced equals the object's whole volume, so the comparison reduces to a contest between the fluid density and the object's average density. Anything denser than the fluid sinks; anything less dense floats. A floating object settles at the depth where the displaced volume produces just enough buoyancy to match its weight.

A worked example

Take a solid object with a volume of V = 0.01 m^3, fully submerged in fresh water. The density of water is rho = 1000 kg/m^3, and g = 9.81 m/s^2. Find the buoyant force, and decide whether the object floats.

Step 1 — identify the displaced volume. The object is fully submerged, so the volume of water it displaces equals its own entire volume:

V_displaced = V = 0.01 m^3

Step 2 — apply Archimedes' principle.

F_b = rho_fluid * g * V_displaced
F_b = 1000 * 9.81 * 0.01
F_b = 98.1 N

So the water pushes up on this object with a force of 98.1 N, regardless of what the object is made of. The buoyant force depends only on the fluid and the displaced volume.

Step 3 — compare with the object's weight. Now the material matters. If the object's own weight is less than 98.1 N — meaning its average density is below 1000 kg/m^3 — the upward buoyant force wins and the object rises until it floats partly out of the water. If its weight exceeds 98.1 N, gravity wins and it sinks. If the weight is exactly 98.1 N, the object hovers at any depth, neutrally buoyant. A weight of 98.1 N corresponds to a mass of 10 kg, so the cut-off here is whether the 0.01 m^3 object masses more or less than 10 kg.

Common mistakes

Using the object's volume when it is only partly submerged. For a floating body, V_displaced is the submerged volume, not the total volume. A log floating with half its bulk above the waterline displaces only the lower half. Plugging in the full volume overestimates the buoyant force.

Bringing the object's density into the buoyant force. F_b depends on the fluid density and the displaced volume — never on what the object is made of. The object's density enters only when you compute its weight for the float-or-sink comparison.

Forgetting that fluid density varies. Seawater is denser than fresh water, so a hull floats slightly higher in the sea than in a river. Warm water is less dense than cold. The same object can float in brine and sink in fresh water if its density sits between the two.

Ignoring trapped air. A sealed hollow object, or a hull with watertight compartments, displaces its full external volume while weighing very little. That is the whole reason a steel ship floats. If the compartments flood, the displaced volume effectively collapses and the buoyancy disappears.

Confusing buoyant force with net force. The buoyant force is only the upward part of the picture. Whether the object actually accelerates depends on F_b minus the weight. A submerged rock still feels 98.1 N of buoyancy per 0.01 m^3 — it just feels even more weight pulling the other way.

Try the interactive NovaSolver calculator

Running one case by hand is straightforward, but watching the float-or-sink threshold shift as you change density and fluid makes the principle stick. The Archimedes Buoyancy Simulator on NovaSolver lets you set the object density, its volume, the submersion fraction, and the fluid — fresh water, seawater, or mercury — and it returns the buoyant force, the weight, the net force, the buoyancy-to-weight ratio, and a clear float, sink, or neutral status.

Related calculators

Buoyancy Calculator — a focused tool for the buoyant force alone, handy when you just need a quick number.
Buoyancy & Stability Simulator — goes a step further to check whether a floating body is stable or will capsize.
Hydrostatic Pressure Simulator — for the depth-dependent pressure that is the physical source of buoyancy in the first place.

You can browse the full set in the fluid mechanics tools hub.

Closing note

Archimedes' principle is one of the oldest results in physics that working engineers still reach for every day. The buoyant force is just the weight of the displaced fluid — rho times g times volume — and an object floats when that force can match its weight. Three ideas carry most of the practical work: use the displaced volume, not the total volume; keep the fluid density and the object density in their separate roles; and remember that trapped air is what lets heavy hulls float. Run the comparison, and the behaviour of almost any body in a fluid becomes predictable.

Faraday's Law of Induction: How a Changing Magnetic Field Makes Voltage

NovaSolver — Fri, 07 Aug 2026 00:17:02 +0000

Spin a bicycle wheel with a dynamo light fitted to the fork and the lamp glows. Pedal harder and it glows brighter. There is no battery anywhere in the system. The light is powered entirely by the rider's legs, converted into electricity by a small magnet sweeping past a coil of copper wire. The same trick, scaled up by a factor of a billion, lights entire cities.

This article explains the physical law behind that conversion — Faraday's law of induction — shows how to compute the induced voltage, and works through a numerical example you can check by hand. It also clears up the sign confusion that trips up almost everyone the first time.

Why this calculation matters

Faraday's law is the operating principle of nearly every machine that generates or transforms electrical power. Generators in power stations, the alternator under a car hood, transformers on utility poles, induction motors, inductive charging pads, metal detectors, and the read heads in older hard drives all rely on the same idea: a changing magnetic flux through a loop of wire drives a voltage around that loop.

If you design or analyze any of these devices, you need to predict that voltage. Size the coil wrong, or misjudge how fast the flux changes, and a generator delivers too little voltage or a transformer overheats. The law also sets hard limits: it tells you why a transformer cannot work on direct current, why a faster-spinning generator produces a higher voltage, and why a coil with more turns is more sensitive. None of that intuition is reliable without the equation behind it.

The core formula

Faraday's law states that the electromotive force, or EMF, induced in a coil equals the negative rate of change of magnetic flux linkage:

EMF = -N * (dPhi/dt)

Here N is the number of turns in the coil, Phi is the magnetic flux through a single turn, and dPhi/dt is how quickly that flux is changing with time. EMF is measured in volts.

The magnetic flux itself is the amount of magnetic field passing through the loop area:

Phi = B * A * cos(theta)

B is the magnetic flux density in tesla, A is the loop area in square metres, and theta is the angle between the field and the normal to the loop. Flux is measured in webers, where one weber equals one tesla times one square metre.

Put the two together and you see there are three independent ways to change the flux and therefore induce a voltage: change the field strength B, change the loop area A, or rotate the loop to change the angle theta. A rotating generator uses the third route; a transformer uses the first.

The minus sign is Lenz's law. It says the induced current flows in whatever direction opposes the change that created it. If the flux through the coil is increasing, the induced current sets up a field that pushes back against the increase. This is not a mathematical decoration — it is conservation of energy. If the induced current reinforced the change instead of opposing it, you would get free energy. For finding the magnitude of the voltage you can drop the sign and use absolute values; keep it only when the direction of current matters.

A worked example

Take a flat coil with N = 200 turns. A magnetic field threads through it, and the flux through one turn falls steadily from 0.05 Wb to 0 over a time interval of 0.1 s. Find the magnitude of the induced EMF.

Step 1 — find the rate of change of flux. The flux drops by 0.05 Wb in 0.1 s, and it does so at a constant rate, so the magnitude of dPhi/dt is simply the change divided by the time:

|dPhi/dt| = 0.05 / 0.1 = 0.5 Wb/s

Step 2 — multiply by the number of turns. Each turn sees the same changing flux, and the turns are in series, so their contributions add:

|EMF| = N * |dPhi/dt| = 200 * 0.5
|EMF| = 100 V

So a coil of 200 turns, with the flux through it collapsing in a tenth of a second, produces 100 volts. That is a substantial voltage from a fairly modest setup, and it points straight at the two design levers: a faster flux change or more turns both raise the EMF. Halve the collapse time to 0.05 s and the voltage doubles to 200 V. Double the turns to 400 and it doubles again. This is exactly the principle every generator and transformer exploits — wind more turns, spin faster, or use a stronger field.

Common mistakes

Expecting a steady field to induce a voltage. A magnet held motionless next to a coil produces nothing, no matter how strong it is. Only a changing flux drives an EMF. This is why transformers need alternating current — direct current gives a constant flux and zero induced voltage on the secondary.

Dropping the number of turns. The flux per turn might be small, but the EMF scales directly with N. Forgetting the factor of N underestimates the voltage of a multi-turn coil by orders of magnitude. The product N times Phi is sometimes called the flux linkage for exactly this reason.

Confusing flux and flux density. B is the field strength in tesla; Phi is the flux in webers, equal to B times area. Faraday's law uses the rate of change of flux, not of field strength alone. Mixing them up scrambles the units.

Reading the minus sign as a real negative voltage. The sign in Faraday's law encodes direction, not a numerical value to subtract. For a magnitude calculation, work with absolute values and use Lenz's law separately to decide which way the current flows.

Assuming the flux changes linearly. The simple formula change-over-time gives the average EMF. If the flux varies sinusoidally, as in a real rotating generator, the instantaneous EMF is the derivative of that sinusoid and the peak value is higher than the average.

Try the interactive NovaSolver calculator

Working one case by hand is useful, but seeing how turns, field strength, and speed trade off in real time builds far better intuition. The Electromagnetic Induction (Faraday's Law) Simulator on NovaSolver animates a magnet moving through a coil: you set the speed mode, the number of turns, the coil area, and the field strength, and it returns the induced EMF, the magnetic flux, and the induced current live, with a chart of flux and EMF against time.

Related calculators

Electromagnetic Induction & Faraday's Law Simulator — a complementary view of induction that emphasises how flux linkage and Lenz's law shape the induced current.
Coulomb Law Simulator — for the electrostatic force between point charges, the starting point of the whole electromagnetic story.
Solenoid Magnetic Field Calculator — to find the field B inside a coil before you feed it into a flux calculation.

You can browse the full set in the electromagnetism tools hub.

Closing note

Faraday's law is compact, but it carries an enormous amount of engineering. The single equation EMF equals minus N times the rate of change of flux explains how a bicycle dynamo, a power-station generator, and a phone charger all do the same job. Three ideas are worth keeping: only a changing flux induces a voltage, the voltage scales with both the number of turns and the speed of the change, and the minus sign is conservation of energy in disguise. Get those right and most induction problems become a matter of identifying what the flux is doing and how fast.