DEV Community: NovaSolver

Induced Drag: Why Producing Lift Always Costs You

NovaSolver — Thu, 23 Jul 2026 12:17:02 +0000

Stand near the end of an airport runway as a heavy jet climbs out and you will sometimes hear a low, rolling rumble that arrives a few seconds after the aircraft has passed. That sound is the trailing vortices — two tightly wound tubes of spinning air streaming off the wingtips. They are not a defect or a sign of a badly built wing. They are the unavoidable signature of a finite wing doing its job: making lift.

Those vortices carry energy, and energy carried away is energy the engines had to supply. That recurring cost has a name — induced drag — and it is one of the few drag terms an aerodynamicist can attack directly through wing geometry. This article explains where induced drag comes from, how to estimate it, and why it shapes everything from glider wings to the winglets on a modern airliner.

Why this calculation matters

Induced drag is the part of total drag that exists only because the wing is producing lift. Unlike parasite drag — skin friction and pressure drag, which a wing pays even at zero lift — induced drag scales with how hard the wing is working. At cruise it might be a quarter of the total. During a slow, heavy climb it can dominate.

That matters for real decisions. Climb performance, ceiling, the speed for best range, the size of the wing, and the case for adding winglets all hinge on the induced-drag term. A designer who underestimates it will promise a climb rate the aircraft cannot deliver. An operator who ignores it will burn fuel flying at the wrong speed. And because induced drag falls as speed rises while parasite drag climbs, the two together set the speed for maximum lift-to-drag ratio — the single most efficient way to fly.

The core formula

A real wing has ends. High-pressure air beneath the wing spills around the tips toward the low-pressure region above, and that spanwise leakage rolls up into the trailing vortices. The vortices tilt the local airflow slightly downward across the whole span — an effect called downwash. Because lift is defined perpendicular to the local flow, that tilt rotates the lift vector slightly rearward. The rearward component is induced drag.

The standard estimate for the induced drag coefficient is:

C_Di = C_L^2 / (pi * AR * e)

Here C_L is the lift coefficient, AR is the wing aspect ratio (span squared divided by wing area), and e is the Oswald efficiency factor. The term e captures how far the real wing departs from an ideal elliptical lift distribution; it is typically between about 0.7 and 0.9, with e = 1 reserved for the perfect elliptical case.

Two features deserve attention. First, induced drag grows with the square of the lift coefficient. Double C_L and you quadruple C_Di. That is why induced drag is worst when the wing is working hardest — low speed, high angle of attack, heavy weight. Second, AR sits in the denominator. A long, slender wing has a high aspect ratio and therefore low induced drag, because the troublesome tips are a smaller fraction of the span.

The total drag coefficient combines both contributions:

C_D = C_D0 + C_Di = C_D0 + C_L^2 / (pi * AR * e)

where C_D0 is the parasite (zero-lift) drag coefficient. The grouping 1/(pi*AR*e) is often written as a single constant K, so C_D = C_D0 + K*C_L^2.

A worked example

Take a finite wing operating at a fairly high lift coefficient, the kind of condition you would see in a climb. The lift coefficient is C_L = 1.2, the aspect ratio is AR = 8, and the Oswald efficiency factor is e = 0.85.

Step 1 — square the lift coefficient.

C_L^2 = 1.2^2 = 1.44

Step 2 — evaluate the denominator.

pi * AR * e = 3.1416 * 8 * 0.85 = 21.36

Step 3 — divide.

C_Di = 1.44 / 21.36 = 0.067

So at this operating point the induced drag coefficient is about 0.067. To put that in context, a clean wing might carry a parasite drag coefficient C_D0 of roughly 0.02 — meaning induced drag here is several times larger than parasite drag. That is the slow-flight regime in a nutshell: the lift-dependent term takes over.

Now notice the leverage. If this wing were stretched to AR = 16 at the same lift coefficient, the denominator would double and C_Di would fall to about 0.034 — roughly half. That single number is why high-performance gliders carry wings so long and thin they look fragile. They spend their lives at modest speeds and high lift coefficients, exactly where induced drag rules, so they buy back efficiency with span.

Common mistakes

Treating induced drag as a fixed quantity. It is not a property of the wing alone. It depends on how much lift the wing is currently producing, which changes with speed, weight, and load factor. The same aircraft has very different induced drag in a gentle cruise and a hard, banked turn.

Confusing it with parasite drag. Parasite drag rises with speed; induced drag falls with speed. Lumping them together hides the trade-off that sets the best-range and best-endurance speeds. Keep the two terms separate until the very last step.

Assuming e is close to 1. The elliptical lift distribution that gives e = 1 is an idealization. Real wings with taper, twist, and a fuselage interrupting the span land well below it. Using e = 1 will make induced drag look better than it is.

Forgetting the load factor in a turn. In a level turn the wing must produce more lift than the aircraft's weight, so C_L rises and induced drag climbs with its square. A 60-degree bank doubles the required lift and roughly quadruples the induced drag — a real reason tight turns bleed energy.

Misreading the aspect-ratio benefit. A higher aspect ratio cuts induced drag but adds structural weight and bending moment at the wing root. The optimum is a balance, not "as slender as possible."

Try the interactive NovaSolver calculator

Working one operating point by hand is straightforward, but the real insight comes from watching how C_Di, total drag, and the lift-to-drag ratio respond as you sweep the inputs. The Lift-Induced Drag Simulator — Wingtip Vortices and L/D Ratio on NovaSolver takes the lift coefficient, aspect ratio, Oswald efficiency, and parasite drag, and returns the induced drag coefficient, the total drag coefficient, the L/D ratio, and the K-factor in real time — alongside a top-down wing planform with the tip-vortex schematic and a drag polar marking your current operating point and the maximum-L/D tangent.

Related calculators

Airfoil Lift Calculator — find the lift coefficient that feeds straight into the induced-drag formula.
Airfoil Simulator — explore how camber and angle of attack shape the pressure field around a section.
Drag Coefficient Calculator — connect drag coefficient to the actual drag force on a body in a flow.

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

Closing note

Induced drag is the price of admission for any wing that does useful work. It is set by three things you can actually influence — how hard the wing is loaded, how slender it is, and how cleanly the lift is distributed across the span — and it punishes overloading harshly because it scales with the square of the lift coefficient. Understand that relationship and a lot of aircraft behaviour stops being mysterious: why gliders look the way they do, why winglets exist, why the best-range speed sits where it does, and why a hard turn always costs you energy. Compute the induced term first, keep it separate from parasite drag, and the rest of the drag picture comes into focus.

Open Channel Flow: Sizing Channels with Manning's Equation

NovaSolver — Thu, 23 Jul 2026 00:17:02 +0000

Stand beside a concrete drainage canal during a storm and the question that matters is simple: will it carry the water, or will it overtop? The water has a free surface open to the atmosphere, it is pulled along by gravity rather than pushed by a pump, and how much of it the channel can carry depends on three things you can measure with a tape and a level. That relationship has had a working answer since the nineteenth century, and it is still the equation civil engineers reach for first.

This article explains open channel flow and Manning's equation, shows how the geometry terms are built, works a full numerical example for a concrete channel, and lists the mistakes that lead to undersized or oversized channels.

Why this calculation matters

Open channel flow is everywhere in civil and environmental engineering: storm drains, irrigation canals, river reaches, sewer mains running partly full, roadside ditches, and spillways. Unlike flow in a pressurized pipe, the driving force is the channel's slope under gravity, and the flow adjusts its own depth to match the discharge. That makes the sizing problem subtle — change the depth and you change the area, the wetted perimeter, and the velocity all at once.

Getting the size right has direct consequences. A channel sized too small overtops and floods during heavy rain. A channel sized too large costs more to build and excavate than it needs to, and may run so shallow and slow that it silts up. Manning's equation is the standard tool for landing between those failures. It is used to find the discharge a channel can carry, the depth that a known discharge will reach, or the slope needed to move a target flow — all from the same relationship.

The core method

Manning's equation gives the steady, uniform flow rate Q in an open channel:

Q = (1/n) * A * R^(2/3) * S^(1/2)

Each term has a clear physical meaning:

n is Manning's roughness coefficient. It captures how much the channel surface resists the flow. Smooth concrete has a low n (around 0.013 to 0.015); earth and gravel channels are rougher (0.022 to 0.035); a weedy natural stream is rougher still.
A is the cross-sectional area of the flowing water.
R is the hydraulic radius, defined as the flow area divided by the wetted perimeter, R = A / P. The wetted perimeter P is the length of channel boundary actually in contact with the water — the bed and the submerged sides, but not the free surface.
S is the channel slope, the drop in elevation per unit length along the channel, a dimensionless number.

The hydraulic radius is the term that does the conceptual work. It expresses how efficiently a cross-section moves water: a large area sharing a short wetted perimeter has a high R and flows well, because less of the water is dragging against a wall. This is why a wide, shallow channel and a narrow, deep one of the same area do not carry the same flow.

Once Q is known, the mean velocity follows immediately from continuity:

V = Q / A

One caution about Manning's equation: as written, it is dimensional. The form above, with the leading coefficient equal to 1, is the SI version, valid when A is in m^2, R is in m, and Q comes out in m^3/s. The US customary form carries a coefficient of 1.49. Always confirm which unit system the constant belongs to.

A worked example

Take a rectangular concrete channel 3 m wide, running 1 m deep, on a slope of S = 0.001. The roughness coefficient for the concrete surface is n = 0.015.

Step 1 — find the flow area. For a rectangular section, area is width times depth:

A = 3 * 1 = 3 m^2

Step 2 — find the wetted perimeter. The water touches the bed (width 3 m) and both side walls (each 1 m deep). The free surface at the top is not counted:

P = 3 + 2 * 1 = 5 m

Step 3 — find the hydraulic radius.

R = A / P = 3 / 5 = 0.6 m

Step 4 — assemble Manning's equation. Substitute n, A, R, and S:

Q = (1/0.015) * 3 * 0.6^(2/3) * 0.001^(1/2)

Evaluate the pieces. The roughness term 1/0.015 = 66.7. The hydraulic radius term 0.6^(2/3) = 0.711. The slope term 0.001^(1/2) = 0.0316. Then:

Q = 66.7 * 3 * 0.711 * 0.0316 = 4.5 m^3/s

The channel carries about 4.5 m^3/s. The mean velocity is Q divided by A, roughly 1.5 m/s — a sensible value for a lined concrete channel, fast enough to resist silting but not so fast as to scour. If a design storm delivered more than 4.5 m^3/s, the engineer would need a steeper slope, a wider section, a smoother lining, or a greater allowed depth.

Common mistakes

Including the free surface in the wetted perimeter. Only the boundary in contact with the water counts — bed and submerged sides. The open top surface is never part of P. Adding it inflates the perimeter and underestimates the flow.

Confusing hydraulic radius with the geometric radius. R = A / P is not the radius of a pipe or a half-circle. For the example channel it is 0.6 m, not 1.5 m. The name is misleading; treat it strictly as area divided by wetted perimeter.

Using the wrong unit constant. The leading coefficient is 1 in SI units and 1.49 in US customary units. Pairing the SI coefficient with feet, or the customary coefficient with metres, throws the discharge off by a large factor.

Picking a roughness coefficient carelessly. Manning's n varies more than people expect — a clean concrete channel and a vegetated earth channel can differ by a factor of two or more, and n changes with maintenance and season. The discharge is directly proportional to 1/n, so a sloppy n choice propagates straight into the answer.

Applying Manning's equation outside its assumptions. The equation describes steady, uniform, fully turbulent flow. It does not capture rapidly varied flow such as a hydraulic jump, backwater curves behind an obstruction, or unsteady flood waves. For those, Manning's equation is a starting point, not the final word.

Try the interactive NovaSolver calculator

Working one cross-section by hand is instructive; comparing shapes and depths quickly is where a tool helps. The Open Channel Flow — Manning's Equation calculator on NovaSolver lets you choose a rectangular, trapezoidal, circular, or triangular channel and adjust the geometry, flow depth, roughness coefficient, and slope, then returns the velocity, flow rate, hydraulic radius, cross-sectional area, Froude number, and flow regime, with live Q-versus-depth curves and a cross-section view.

Related calculators

Pipe pressure drop calculator — for the closed-conduit counterpart, where flow is pressure-driven rather than gravity-driven.
Reynolds number calculator — to confirm a channel flow is turbulent, the regime Manning's equation assumes.
Pipe flow calculator — for relating flow rate, velocity, and diameter in full-flowing pipes.

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

Closing note

Manning's equation is a small calculation that decides whether a channel does its job. It ties together three things an engineer can measure — roughness, geometry through the hydraulic radius, and slope — and returns a discharge you can defend. Count the wetted perimeter correctly, keep the unit constant consistent with your length units, choose the roughness coefficient with care, and remember the equation assumes steady uniform flow. With those four habits, sizing a channel becomes a clear, repeatable calculation rather than a guess.

Airfoil Lift: How a Wing Holds an Aircraft in the Air

NovaSolver — Wed, 22 Jul 2026 12:17:02 +0000

A fully loaded airliner weighing hundreds of tonnes climbs away from a runway on nothing but moving air. There is no trick and no exotic physics — just a wing of the right shape passing through the air fast enough. The force that lifts it can be written in a single equation, and once you can read that equation you can answer practical questions: how fast must this wing go to fly, how much load can it carry, what happens if the air thins out at altitude.

This article explains where the lift equation comes from, what each term means, works a full numerical example for a small aircraft, and flags the mistakes that creep in when people apply it.

Why this calculation matters

Lift is the force that opposes weight. In steady level flight the two balance exactly, so the lift equation is also the equation that decides whether an aircraft can fly at all, and at what speed. It is the first calculation in wing sizing, in setting stall and approach speeds, and in estimating how payload trades against runway length.

The same equation governs anything that uses a wing-like surface: wind turbine blades, propellers, helicopter rotors, racing-car wings that push down instead of up, and sails. It also explains why aircraft behave differently with altitude and temperature. Lift depends directly on air density, and density falls as you climb or as the air warms. A wing that lifts comfortably at sea level on a cold morning may struggle from a hot, high-altitude runway. None of that is intuitive until you have the equation in front of you.

The core formula

Lift L produced by a wing is given by:

L = 0.5 * rho * V^2 * C_L * A

where rho is the air density, V is the airspeed (the speed of the wing relative to the air), C_L is the lift coefficient, and A is the wing planform area — the area of the wing seen from directly above.

The group 0.5 * rho * V^2 is the dynamic pressure, often written q. It is the pressure the moving air would exert if brought to rest, and it carries the strong dependence on speed: lift scales with the square of airspeed, so flying twice as fast at the same attitude produces four times the lift. That single fact explains why takeoff and landing happen at carefully chosen speeds and why small speed errors matter near stall.

The lift coefficient C_L packages everything about the wing's shape and orientation into one dimensionless number. It depends mainly on the airfoil profile, the wing's aspect ratio, and the angle of attack — the angle between the wing chord and the oncoming air. For a typical airfoil, C_L rises roughly linearly with angle of attack up to a point, then the flow separates from the upper surface and the wing stalls: C_L drops sharply even though the nose is still rising. The peak value of C_L before stall is one of the defining numbers of a wing.

Two readings are worth keeping. First, lift can be increased by flying faster, by flying in denser air, by enlarging the wing, or by raising C_L through angle of attack or flaps. Second, C_L is not a property you can push without limit — the stall ceiling is real and is one of the hardest constraints in flight.

A worked example

Take a light aircraft in cruising flight. The air density is rho = 1.2 kg/m^3, the airspeed is V = 60 m/s, the wing planform area is A = 16 m^2, and the wing is operating at a lift coefficient C_L = 1.2.

Step 1 — find the dynamic pressure. This is the speed-dependent part:

0.5 * rho * V^2 = 0.5 * 1.2 * 60^2 = 0.5 * 1.2 * 3600 = 2160 Pa

Step 2 — multiply by the lift coefficient and wing area.

L = 2160 * 1.2 * 16
L = 41,472 N

So the wing produces about 41.5 kN of lift.

Step 3 — convert lift to the mass it supports. In steady level flight lift equals weight, so divide by g = 9.81 m/s^2:

m = L / g = 41,472 / 9.81 = 4,227 kg

A wing producing 41,472 N of lift holds up a mass of about 4,227 kg — a light aircraft at cruise. The example is worth re-reading as a sanity check: a 16 m^2 wing, an ordinary cruise speed, and a moderate lift coefficient combine to support several tonnes. Lift is not a marginal force; a properly sized wing produces it in abundance.

Common mistakes

Forgetting the factor of one-half. The dynamic pressure is 0.5 * rho * V^2, not rho * V^2. Dropping the one-half doubles the lift and is one of the most common slips in this calculation.

Using the wrong area. A is the planform area — the wing seen from above — not the wetted (total surface) area and not a cross-section. Mixing these up can change the answer by a large factor.

Treating C_L as a fixed constant. The lift coefficient depends on angle of attack and changes continuously as the aircraft maneuvers. It also cannot exceed the stall value. Plugging in a high C_L that the wing cannot actually reach gives a lift number the wing will never produce.

Ignoring the change in air density. Density falls with altitude and rises in cold air. The same wing at the same speed and attitude produces less lift at altitude than at sea level. Using sea-level density for a high-altitude case overstates the lift.

Confusing airspeed with ground speed. Lift depends on speed relative to the air, not relative to the ground. A strong headwind or tailwind changes ground speed without changing lift; only the airspeed term V belongs in the equation.

Try the interactive NovaSolver calculator

The arithmetic is quick, but the interesting behaviour lives in how angle of attack, airfoil shape, speed, and air density interact — and especially in how the wing approaches stall. The Airfoil Lift Calculator — CL, CD & L/D with Thin Airfoil Theory on NovaSolver lets you pick a NACA airfoil profile and adjust angle of attack, chord, span, airspeed, and air density, then returns the lift coefficient, drag coefficient, lift and drag forces, lift-to-drag ratio, and dynamic pressure, with a live polar curve that shows the stall behaviour directly.

Related calculators

Airfoil calculator — to explore airfoil geometry and pressure distribution before settling on a profile.
Lift-induced drag calculator — to see the drag penalty that producing lift inevitably brings, which depends on aspect ratio.
Drag coefficient calculator — to estimate the total drag force a body or wing must overcome.

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

Closing note

The lift equation is one of the most rewarding formulas in engineering: four quantities you can measure, combined to give the force that keeps an aircraft aloft. The takeaways are compact. Lift grows with the square of airspeed, falls with thinning air, scales with wing area, and is shaped by a lift coefficient that has a hard ceiling at stall. Keep the factor of one-half, use the planform area, respect the stall limit, and the equation will tell you honestly what a wing can and cannot do.

Welding Heat Input: The Number That Decides Weld Quality

NovaSolver — Wed, 22 Jul 2026 00:17:01 +0000

A welder lays down two beads on the same steel plate with the same electrode. One bead cools fast and turns hard and brittle; the other cools slowly and stays tough. Nothing about the steel changed. What changed was how much energy the arc poured into each millimetre of weld — and that single quantity, welding heat input, quietly governs almost everything that happens in the joint afterward.

This article explains what heat input is, how to calculate it from the parameters a welder actually controls, works a full numerical example, and points out the mistakes that cause heat input to be misread on a procedure sheet.

Why this calculation matters

Heat input sets the thermal history of a weld. A high heat input produces a large molten pool and a slow cooling rate. A low heat input does the opposite. Because the mechanical properties of a welded joint depend far more on cooling rate than on the filler alloy alone, heat input ends up controlling hardness, toughness, grain size in the heat-affected zone (HAZ), and the residual stress and distortion left behind.

This is not an academic concern. Welding procedure specifications (WPS) for structural steel, pressure vessels, and pipelines almost always state a heat input range, not just a current and voltage. Stay below the lower limit and the HAZ can cool fast enough to form brittle, crack-prone martensite. Run above the upper limit and the weld metal grows coarse grains and loses toughness, while distortion increases. The whole point of qualifying a procedure is to land inside a window — and heat input is the coordinate that defines that window.

The core formula

Heat input H is the arc energy delivered per unit length of weld. The arc supplies electrical power equal to voltage times current; the travel speed determines how that power is spread along the joint:

H = (eta * V * I) / v

where eta is the arc (thermal) efficiency, V is the arc voltage in volts, I is the welding current in amperes, and v is the travel speed. With current in A, voltage in V, and speed in mm/s, H comes out in joules per millimetre (J/mm). Multiply by 1000 to read kJ/mm, the unit most procedure sheets use.

The efficiency eta accounts for the fact that not all electrical energy enters the workpiece — some is lost as radiation, spatter, and convection from the arc. It depends on the process:

SAW  (submerged arc)   eta ~ 0.90 - 0.99
SMAW (stick)           eta ~ 0.70 - 0.80
GMAW / MIG             eta ~ 0.75 - 0.85
GTAW / TIG             eta ~ 0.55 - 0.70
Laser                  eta ~ 0.40 - 0.50

Two relationships are worth internalizing. Heat input rises with voltage and current — more arc power — and falls as travel speed rises, because the same power is smeared over a longer bead. A welder who speeds up to "finish faster" is also, whether they intend to or not, reducing heat input and accelerating the cooling rate.

A useful companion concept is the cooling rate. A higher heat input slows cooling through the critical temperature range (often quantified as the time to cool from 800 to 500 C, the t8/5 time). Slow cooling favours softer, tougher microstructures; fast cooling favours hard, brittle ones. Heat input is the lever the welder has over that rate.

A worked example

Consider an arc weld run with the following parameters: arc efficiency eta = 0.8, arc voltage V = 25 V, welding current I = 200 A, and travel speed v = 5 mm/s.

Step 1 — assemble the arc power term. The electrical power delivered to the arc is V times I:

V * I = 25 * 200 = 5000 W

Step 2 — apply the efficiency. Only a fraction eta of that power enters the workpiece:

eta * V * I = 0.8 * 5000 = 4000 W

Step 3 — divide by travel speed. Spreading 4000 W of effective power over a bead advancing at 5 mm/s gives the energy per millimetre:

H = 4000 / 5 = 800 J/mm

So this weld receives a heat input of 800 J/mm, or 0.8 kJ/mm. That is a moderate value — typical for structural fillet welding on medium-thickness steel.

The practical reading is the part that matters. A higher heat input means a larger weld pool and a slower cooling rate, which shifts the microstructure and the residual stress of the joint. If a procedure called for 0.8 to 1.2 kJ/mm, this weld sits at the lower edge of the window. Slowing the travel speed to 3.3 mm/s would push H to about 1.2 kJ/mm — the upper edge — without touching the current or voltage at all. That is the kind of trade a welding engineer makes routinely.

Common mistakes

Forgetting the efficiency factor. Some references quote "heat input" as simply V times I divided by v, with no eta. That is the arc energy, not the heat delivered to the plate. Mixing the two overstates the real input — by 25 % for stick welding, by more than half for laser. Always check whether a stated value includes eta.

Mismatched speed units. Travel speed appears on equipment in mm/s, mm/min, or in/min. Use mm/s with current in A and voltage in V to get J/mm directly. Plugging mm/min into the same formula inflates the answer by a factor of 60.

Treating voltage and current as independent of the process. Efficiency is tied to the process, not chosen freely. A TIG weld and a submerged-arc weld at the same V and I do not deliver the same heat. Pick eta from the process before computing anything.

Assuming more heat input is always safer. It is not. High heat input lowers hardness and cracking risk in many steels, but it also coarsens grains, lowers HAZ toughness, and increases distortion. The goal is a qualified range, not a maximum.

Ignoring plate thickness and joint geometry. The same heat input cools far faster in thick plate than in thin sheet, because thick sections conduct heat away in three dimensions. Heat input is one input to cooling rate, not the whole story.

Try the interactive NovaSolver calculator

Running the arithmetic once is straightforward; seeing how current, voltage, travel speed, and process efficiency trade off against cooling rate is where a tool earns its place. The Welding Heat Input & Cooling Rate Calculator (Rosenthal Solution) on NovaSolver lets you pick the welding process — GMAW, GTAW, SMAW, SAW, or laser — and adjust current, voltage, travel speed, thermal efficiency, plate thickness, material, and preheat temperature, then returns the heat input along with the peak temperature distribution, HAZ width, and cooling rate from the Rosenthal analytical solution.

Related calculators

Fillet weld calculator — to size the throat and leg of a fillet weld once you know the process and load.
Weld joint strength calculator — to check whether a welded joint carries its design load with adequate margin.
Welding residual stress calculator — to estimate the locked-in stress that heat input leaves behind after the weld cools.

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

Closing note

Welding heat input is a compact calculation with outsized consequences. It is just effective arc power divided by travel speed, but that ratio fixes the size of the weld pool, the speed of cooling, the microstructure of the HAZ, and the distortion of the finished part. Compute it with the right efficiency, keep the speed units consistent, and treat it as a qualified range rather than a single target. Get the heat input right and most of the metallurgy that follows takes care of itself.

Casting Solidification Time: Why a Thick Section Cools So Much Slower

NovaSolver — Tue, 21 Jul 2026 12:17:03 +0000

Pour two castings from the same ladle of the same metal into the same kind of sand mould. One is a thin plate, the other a chunky block. The plate is solid before you have walked back to the furnace; the block is still mushy in the middle minutes later. The metal did not change. The mould did not change. What changed is the geometry — specifically, how much surface the casting offers for its volume of heat to escape through.

That single geometric idea is the heart of foundry practice. Get it right and you can predict which section freezes last, place a riser to feed it, and produce a sound casting. Get it wrong and the last region to solidify shrinks against fully frozen metal around it, leaving a void. This article explains Chvorinov's rule, the casting modulus it depends on, and how to turn both into a solidification time.

Why this calculation matters

A casting does not solidify all at once. Thin sections freeze first, thick sections last, and liquid metal shrinks as it turns solid. Unless that shrinkage is fed with extra liquid from a reservoir, the last region to freeze ends up with a shrinkage cavity — porosity that can scrap the part or seed a fatigue crack in service.

The whole craft of riser design rests on one principle: the riser must stay liquid longer than the casting section it feeds. To apply that principle you need to know, quantitatively, how long different sections take to solidify. Chvorinov's rule gives you that number from geometry alone, before any metal is poured. It tells you which section is the thermal bottleneck, how big a riser must be to outlast it, and where to chill or pad the casting to steer the freezing front. It also explains, in one tidy equation, why a heavy boss or a thick flange is so often the source of trouble.

The core method

Heat leaves a solidifying casting through its surface. The more volume a casting holds relative to its cooling surface area, the longer that heat takes to escape. Chvorinov captured this in a compact rule: solidification time is proportional to the square of the volume-to-surface-area ratio.

t = B * (V / A)^2

Here t is the solidification time, V is the casting volume, A is the surface area through which heat is lost, and B is the mould constant. The ratio V/A appears so often in foundry work that it has its own name — the casting modulus, usually written M:

M = V / A

The modulus has units of length and is, loosely, a measure of section thickness: a thin wall has a small modulus, a chunky block a large one. Because solidification time grows with the square of the modulus, doubling the modulus quadruples the freezing time. That square law is why thick sections cool so disproportionately slowly.

The mould constant B bundles together everything about the mould and the metal that is not geometry — the mould material's ability to absorb heat, the metal's latent heat of fusion, the pouring superheat, and the freezing temperature. It is found by experiment for a given metal-and-mould combination and has units of time per length squared, for example min/cm^2. The practical power of the rule is this: compute the modulus of the casting section and the modulus of a candidate riser, and require the riser's modulus to be the larger of the two. A riser with a higher modulus freezes later and can feed the casting until the casting itself is solid.

A worked example

Take a casting with a volume V = 1000 cm^3 cooling through a surface area A = 600 cm^2, poured into a mould whose constant is B = 2 min/cm^2.

Step 1 — compute the casting modulus. The modulus is the volume-to-area ratio:

M = V / A = 1000 / 600 = 1.67 cm

Step 2 — square the modulus. Chvorinov's rule needs the square of the modulus:

M^2 = (1.67)^2 = 2.78 cm^2

Step 3 — apply the mould constant. Multiply by B to get the solidification time:

t = B * M^2
t = 2 * 2.78
t = 5.6 min

So this casting takes about 5.6 minutes to solidify. The value of the number is comparative. Work out the modulus of every distinct section, and the one with the largest modulus freezes last — that is where shrinkage will concentrate and where a riser must be placed. A riser feeding this section must be sized so its own modulus exceeds 1.67 cm; otherwise it freezes first and feeds nothing. A thin-walled section of the same casting, with a much smaller modulus, would solidify in a small fraction of this time and need no feeding at all.

Common mistakes

Counting non-cooling surfaces in the area. Only surfaces through which heat actually escapes belong in A. Where two casting sections join, or where a riser meets the casting, no heat crosses — those interfaces are not cooling area. Including them inflates A, shrinks the modulus, and underestimates the freezing time.

Treating solidification time as linear in size. It is not. Time scales with the square of the modulus, so a section twice as thick freezes roughly four times more slowly. Linear intuition badly underestimates how stubborn a heavy section is.

Forgetting that the mould constant is specific. B is tied to one particular metal and one particular mould material. A value measured for sand-cast aluminum does not transfer to a steel casting in a different mould. Using a borrowed B can throw the time off by a wide margin.

Sizing a riser by volume instead of modulus. A riser that merely holds enough metal can still freeze before the casting does. The governing requirement is that the riser's modulus exceed the casting section's modulus, so the riser stays liquid longest.

Ignoring chills and section transitions. A chill placed against a thick section raises the local heat loss and shortens its freezing time, deliberately reshaping the modulus map. Sharp transitions between thick and thin sections create hot spots that a uniform-wall assumption will miss.

Try the interactive NovaSolver calculator

Working the square of the modulus by hand is fine once, but seeing how volume and cooling area trade off against freezing time is far more instructive. The Casting Solidification Chvorinov Simulator on NovaSolver lets you set the casting volume, the cooling surface area, the mould constant, and the exponent, and it returns the solidification time, the casting modulus, a cooling index, and a riser-margin guide, with linked geometry, timeline, and sensitivity views so you can watch how a chunkier casting freezes more slowly.

Related calculators

CNC Machining — Cutting Force & Tool Life Calculator — for the downstream step, machining a solidified casting to its finished dimensions.
Weld Heat Input Calculator — for another process governed by how fast a molten pool loses heat to the surrounding metal.
Welding Residual Stress Calculator — to see how uneven cooling, the same physics behind casting shrinkage, locks stress into a part.

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

Closing note

Chvorinov's rule is one of those results that pays back its simplicity many times over. Solidification time depends on the square of the volume-to-surface-area ratio, and that one square law explains why heavy sections are slow to freeze, why they collect shrinkage porosity, and why a riser must always have the larger modulus. Compute the modulus of every section, find the one that freezes last, and feed it deliberately — and a casting that might have hidden a void in its thickest region comes out sound instead.

CNC Machining Cutting Speed: Turning Spindle RPM Into Real Surface Speed

NovaSolver — Tue, 21 Jul 2026 00:17:02 +0000

A new operator sets a lathe to 600 rpm, runs the job, and the tool edge is glazed and chipped within minutes. The next day, the same part on the same machine at the same rpm cuts cleanly for an hour. The only difference: a different bar diameter. The spindle speed never told the whole story, because the spindle speed is not what the tool feels at the cutting edge.

What the tool feels is cutting speed — the rate at which the workpiece surface sweeps past the cutting edge. It depends on diameter as well as rpm, and it is the quantity that tool-life data, surface-finish guidance, and recommended-speed tables are all written against. This article walks through how to convert spindle speed into cutting speed, how to combine it with feed and depth of cut into a material removal rate, and where the conversion most often goes wrong.

Why this calculation matters

Cutting speed is the master variable of any machining process. Tool wear is dominated by the temperature at the cutting edge, and that temperature climbs steeply with cutting speed. Tool-life models such as Taylor's equation are expressed directly in terms of cutting speed, not spindle rpm. A handbook that recommends 90 m/min for carbon steel with a carbide tool is giving you a surface speed; it cannot give you an rpm, because the right rpm depends on the diameter you happen to be cutting.

That diameter dependence is the practical trap. The same 600 rpm produces a gentle 47 m/min on a 25 mm bar and an aggressive 188 m/min on a 100 mm bar. Run the large bar at small-bar rpm and the edge overheats; run the small bar at large-bar rpm and you leave productivity on the table. Cutting speed is also the bridge to material removal rate, which sets cycle time and the power the spindle must deliver. Get the speed conversion right and the rest of the process plan — tool life, finish, cost per part — rests on a solid foundation.

The core formula

In a turning operation, cutting speed is the tangential speed of the workpiece surface at the point of contact. A point on the outer surface travels one circumference, pi times the diameter, every revolution, so:

V = pi * D * N

Here V is the cutting speed, D is the workpiece diameter, and N is the spindle speed in revolutions per minute. The standard industrial unit for V is metres per minute, so keep D in metres when you want V in m/min directly. The single most important habit is to use the actual diameter being cut — in turning, that is the current bar diameter, which shrinks as material comes off.

Cutting speed alone does not tell you how fast metal leaves the part. For that you need the material removal rate, the volume of material cut away per unit time. For turning it is the product of three quantities:

MRR = V * f * d

V is the cutting speed, f is the feed per revolution, and d is the depth of cut. The geometric picture is simple: each revolution the tool sweeps a thin ribbon of metal whose width is the depth of cut and whose thickness is the feed, and the cutting speed sets how fast that ribbon is produced. MRR is what determines cycle time and, together with the material's specific cutting energy, the spindle power the job demands. Keep the units consistent — if V is in mm/min, then f in mm/rev and d in mm give MRR in mm^3/min.

A worked example

Take a turning job on a bar of diameter D = 50 mm running at a spindle speed N = 600 rpm.

Step 1 — convert rpm to cutting speed. Put the diameter in metres, 0.050 m, and apply V = pi*D*N:

V = pi * 0.050 * 600
V = 94.2 m/min

So the workpiece surface is sweeping past the tool at about 94 m/min. That is the number to check against a recommended-speed table for the material and tool grade — not the 600 on the dial.

Step 2 — set the feed and depth of cut. Use a feed of f = 0.2 mm/rev and a depth of cut of d = 2 mm.

Step 3 — compute the material removal rate. MRR needs consistent units, so express the cutting speed in mm/min: 94.2 m/min is 94,200 mm/min. Then:

MRR = V * f * d
MRR = 94,200 * 0.2 * 2
MRR = 37,680 mm^3/min

Converting to the more readable unit, that is about 37.7 cm^3/min of steel removed every minute. With the part volume and this rate you can estimate roughing time directly, and with the material's specific cutting energy you can estimate the spindle power the cut will draw.

Common mistakes

Treating spindle rpm as cutting speed. They are proportional only at a fixed diameter. The same rpm gives a different cutting speed on every bar size, and on a facing cut the speed varies continuously as the tool moves toward the centre. Always convert through V = pi*D*N.

Using the wrong diameter. In turning, use the diameter being cut, not the chuck size or the nominal stock size. In milling the roles swap — it is the cutter diameter that matters, not the workpiece. Picking the wrong D scales the cutting speed by exactly the wrong factor.

Mixing units in the MRR formula. A frequent slip is leaving V in m/min while feed and depth are in millimetres. The factor-of-1000 mismatch makes MRR look a thousand times too small. Convert everything to one length unit before multiplying.

Forgetting that facing speed is not constant. When facing toward the centre at fixed rpm, the cutting speed falls to zero at the axis. Constant-surface-speed control on the lathe exists precisely to hold V steady by raising rpm as the diameter shrinks.

Ignoring the power and tool-life consequences of a high MRR. A large removal rate is attractive for cycle time but draws more spindle power and shortens tool life. The fastest MRR your math allows is not always the one the machine, the fixture, or the budget can sustain.

Try the interactive NovaSolver calculator

Running the conversion once is straightforward; building a feel for how diameter, rpm, feed, and depth trade off against tool life and power is the harder part. The CNC Machining — Cutting Force & Tool Life Calculator on NovaSolver covers turning, milling, and drilling, and reports cutting speed, cutting force from the Merchant model, material removal rate, Taylor tool life, theoretical surface roughness, and spindle power, with built-in material data for aluminum, steel, stainless, titanium, and cast iron so you can compare operating points side by side.

Related calculators

Casting Solidification (Chvorinov) Simulator — for the upstream step, estimating how long a cast blank takes to solidify before it reaches the machine.
Weld Heat Input Calculator — for the joining side of manufacturing, where travel speed plays a role much like feed rate.
Weld Joint Strength Calculator — to check that a machined and welded assembly carries its load.

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

Closing note

Cutting speed is the quiet variable behind every machining decision. It is a one-line conversion — multiply pi, diameter, and rpm — but it is the speed your tool actually experiences, the speed your handbook data was written for, and the speed that sets your tool life. Combine it with feed and depth of cut and you have the material removal rate that drives cycle time and power. Convert rpm to cutting speed first, use the diameter that is really being cut, keep your units honest, and the rest of the process plan follows.

Lead-Lag Compensators: Buying Phase Margin Back for a Sluggish Loop

NovaSolver — Mon, 20 Jul 2026 12:17:02 +0000

A position servo that hunts around its target, an autopilot that wallows after every gust, a power supply that rings before it settles — these symptoms often trace back to the same root cause. The loop has enough gain to be fast, but not enough phase margin to be calm. Push the gain up for speed and the system edges toward oscillation; back the gain off for stability and the response turns sluggish. You appear to be stuck choosing between two things you wanted at once.

The lead-lag compensator is the classical escape from that bind. Instead of trading gain against stability, it reshapes the loop's frequency response so you can keep the gain you need and recover the phase margin you lost. This article explains how a lead compensator adds phase, where it adds it, and how much you can realistically expect from one stage.

Why this calculation matters

Phase margin is the single most useful health indicator for a feedback loop. It measures how much extra phase lag the loop can absorb at the gain crossover frequency before the closed loop goes unstable. A loop with 45 to 60 degrees of phase margin is usually well damped; one with 10 to 20 degrees rings badly; one near zero is on the edge of sustained oscillation.

The trouble is that almost everything you add to a real loop eats phase margin. Actuator dynamics, sensor filters, transport delay, and the integrator you wanted for zero steady-state error all contribute lag. By the time the loop is doing its job, the phase margin can be uncomfortably thin. A lead compensator lets you inject phase lead back into the loop, concentrated near the crossover frequency where it counts, without having to remove the very elements that caused the lag. Knowing how much phase a single stage can supply tells you immediately whether one compensator will fix the problem or whether you need two.

The core formula

A lead compensator is a first-order transfer function with a zero and a pole, written as:

C(s) = (1 + s*T) / (1 + s*alpha*T),   with alpha < 1

The zero sits at frequency 1/T and the pole sits at the higher frequency 1/(alpha*T). Because the zero comes first, the compensator adds positive phase — phase lead — across the band between the two break frequencies. The parameter alpha controls how far apart the zero and pole are spaced, and that spacing sets the strength of the compensator.

The maximum phase lead a single stage can produce depends only on alpha:

sin(phi_max) = (1 - alpha) / (1 + alpha)

The smaller alpha is, the wider the pole-zero gap and the larger phi_max. That phase peak does not occur just anywhere — it lands at the geometric mean of the two break frequencies:

omega_m = 1 / (T * sqrt(alpha))

This is the key design lever. You choose alpha to get the phase lead you need, then choose T so that omega_m lands exactly on the gain crossover frequency, where the phase margin is measured. Get those two choices right and the lead appears precisely where the loop is weakest. One caveat: the same pole-zero pair also raises the loop gain by a factor of 1/alpha at high frequency, which nudges the crossover frequency upward. Good design accounts for that shift rather than ignoring it.

A worked example

Suppose a loop needs more phase margin and you decide to use an aggressive single lead stage with alpha = 0.1.

Step 1 — compute the maximum phase lead. Substitute alpha into the phase formula:

sin(phi_max) = (1 - 0.1) / (1 + 0.1)
sin(phi_max) = 0.9 / 1.1
sin(phi_max) = 0.818

Step 2 — solve for the angle.

phi_max = arcsin(0.818) = 54.9 degrees

So a single lead stage with a tenfold pole-zero separation contributes about 55 degrees of phase lead at its peak frequency. That is a large amount. If a loop has a phase margin of 20 degrees and needs to reach 55 to 60 degrees, one well-placed lead stage can close most or all of that gap on its own.

Step 3 — place the peak. Once alpha is fixed, choose T from omega_m = 1/(T*sqrt(alpha)) so the peak lands at the gain crossover frequency. With alpha = 0.1, sqrt(alpha) is about 0.316, so T = 1/(omega_m * 0.316). Pick omega_m equal to the desired crossover, solve for T, and the compensator's phase boost arrives exactly where the phase margin is measured.

A practical note on limits: alpha values much below 0.1 are rarely used. As alpha shrinks, the high-frequency gain boost of 1/alpha grows, which amplifies sensor noise and can push crossover so high that unmodeled dynamics intrude. If you need more than about 55 to 60 degrees of lead, two cascaded stages are cleaner than one extreme stage.

Common mistakes

Placing the phase peak away from crossover. The lead is only useful if omega_m coincides with the gain crossover frequency. A compensator tuned so its peak lands a decade away delivers almost none of its phase benefit where the margin is actually evaluated.

Forgetting the gain that comes with the phase. A lead stage is not phase-only. It raises high-frequency gain by 1/alpha, which moves the crossover frequency to a higher value. If you size T for the old crossover and ignore the shift, the phase peak ends up below the new crossover.

Chasing tiny alpha values. It is tempting to drop alpha toward 0.01 to squeeze out more lead, but the high-frequency gain boost and noise amplification grow just as fast. Past roughly 55 to 60 degrees from one stage, cascading two moderate stages is the better engineering choice.

Confusing lead with lag. A lag compensator has its pole below its zero and adds phase lag, not lead. It is used to raise low-frequency gain for steady-state accuracy, not to improve phase margin. Mixing up which element you need is a common source of wasted tuning effort.

Ignoring noise. Lead action is differentiating in character. It sharpens the response but also amplifies measurement noise above the zero frequency. A loop that looks great on a Bode plot can still be unusable if the control signal is buried in noise.

Try the interactive NovaSolver calculator

Seeing how the zero and pole move the magnitude and phase curves is far more instructive than working the trigonometry once. The Bode Lead Lag Compensator Simulator on NovaSolver links the magnitude, phase, and pole-zero views in one place — adjust the gain, the zero, the pole, and the crossover frequency, and it reports the phase contribution at crossover, the gain at crossover, the pole-zero separation, and a margin estimate, so you can watch the stability picture change as you spread or close the pole-zero gap.

Related calculators

Bode Plot Generator — to see the full open-loop magnitude and phase response before and after you add a compensator.
PID Controller Simulator — for the time-domain alternative, where derivative action plays a role similar to lead.
Root Locus Plotter — to track how adding a compensator zero and pole pulls the closed-loop poles across the s-plane.

You can explore the rest in the controls and frequency response tools hub.

Closing note

A lead compensator is a small, elegant tool with a clear job: put phase lead where the loop needs it. The design reduces to two decisions — pick alpha for the amount of lead, pick T so the peak lands at crossover — and a single stage can supply roughly 55 degrees, often enough to rescue a marginal loop. Keep the gain boost and the noise penalty in view, place the peak deliberately, and the lead-lag compensator turns the false choice between speed and stability back into a design you can actually tune.

The Nyquist Stability Criterion: Knowing a Loop Is Stable Before You Close It

NovaSolver — Mon, 20 Jul 2026 00:17:02 +0000

Anyone who has held a microphone too close to its own speaker knows the sound: a quiet hum that swells into a piercing howl within a second. Nothing about the room changed in that second. What changed is that a signal found a path back to its own input with enough strength and the wrong timing, and the loop fed on itself. That runaway is instability, and the unsettling part is that it was latent the whole time — the system looked fine until it suddenly was not.

The Nyquist stability criterion exists to remove that surprise. It lets you take the open-loop behaviour of a system — the loop measured or modelled with the feedback path broken — and decide, before ever closing the loop, whether the closed loop will be stable. This article explains the idea, shows how to read the stability margins it produces, and works a concrete example.

Why this calculation matters

Feedback is everywhere a system has to hold a target: cruise control, a voltage regulator, a drone holding altitude, a furnace holding temperature. Feedback is also what makes those systems capable of instability. Close a loop carelessly and you do not get a slightly worse response — you can get sustained oscillation or a divergent one.

The deep convenience of the Nyquist criterion is that it answers a closed-loop question using open-loop data. The open-loop response is the easy thing to obtain: you can compute it from the plant and controller models, or measure it directly by injecting a signal and breaking the loop. From that one curve, Nyquist tells you the fate of the closed loop. Just as important, it does not give a bare pass or fail — it tells you how close to instability you are, through the gain and phase margins. Those margins are what separate a robust design that tolerates real-world variation from a fragile one that was only ever stable on paper.

The core method

The Nyquist plot is the open-loop frequency response drawn as a single curve in the complex plane. As frequency omega sweeps from zero upward, the open-loop transfer function L(j*omega) traces a path; its distance and angle from the origin are the loop gain and phase at each frequency. The criterion compares how that curve relates to one special location — the point at -1 on the real axis, which represents a gain of exactly 1 with a phase of exactly -180 degrees. That is the condition for a signal to return to the input unchanged and self-sustaining.

For the common case of a stable open-loop system, the rule reduces to something simple: the closed loop is stable if the Nyquist curve does not encircle the -1 point. How far the curve stays from -1 is the margin, and we read that margin at two specific frequencies.

The phase margin is measured at the gain crossover frequency — the frequency where the open-loop gain passes through 1, or 0 dB. It is the extra phase lag the loop could absorb at that frequency before the phase reaches -180 degrees:

phase margin = 180 + (open-loop phase at gain crossover)

The gain margin is measured at the phase crossover frequency — the frequency where the open-loop phase passes through -180 degrees. It is how much the gain could rise before it reaches 0 dB:

gain margin (dB) = 0 - (open-loop gain in dB at phase crossover)

In words: phase margin is your buffer against extra delay, and gain margin is your buffer against extra gain. A healthy loop has both.

A worked example

Suppose you have measured the open-loop frequency response of a feedback loop and want to read its stability margins.

Step 1 — find the gain crossover and read the phase margin. Locate the frequency where the open-loop gain is 0 dB. Read the phase at that frequency: it is -135 degrees. The phase margin is the gap to -180 degrees:

phase margin = 180 - 135 = 45 degrees

The loop could tolerate an additional 45 degrees of phase lag — perhaps from an unmodelled delay — before the phase at gain crossover hits -180 degrees.

Step 2 — find the phase crossover and read the gain margin. Locate the frequency where the open-loop phase is -180 degrees. Read the gain there: it is -8 dB. The gain margin is how far that sits below 0 dB:

gain margin = 0 - (-8) = 8 dB

The loop gain could rise by 8 dB — a factor of about 2.5 — before the gain at phase crossover reaches unity.

Step 3 — judge the result. A phase margin near 45 degrees and a gain margin in the 6 to 10 dB band are the textbook signature of a well-damped, robustly stable loop. This design has 45 degrees and 8 dB, so it sits comfortably in that target zone. It will settle quickly without ringing excessively, and it has real headroom against the gain drift and extra lag that any physical system eventually develops.

Common mistakes

Reading margins off the closed-loop response. Gain and phase margins are defined on the open-loop curve. Computing them from the closed-loop frequency response confuses two different functions and gives numbers that mean nothing.

Assuming a stable open-loop system makes the simple rule always apply. The "no encirclement of -1" shortcut holds when the open-loop transfer function has no poles in the right half-plane. If the plant is itself unstable, the full criterion counts encirclements against those poles, and the closed loop may actually need a specific number of encirclements to be stable.

Trusting one margin alone. A loop can show a generous gain margin and a poor phase margin, or the reverse. Conditionally stable systems can even cross 0 dB more than once. Always read both margins, and on a tricky plot check the whole curve, not just the two crossover points.

Treating a large phase margin as always better. Very high phase margin often means a sluggish, overdamped response. Margins around 45 to 60 degrees usually balance speed against robustness; chasing 80 degrees can cost you bandwidth you wanted.

Forgetting pure time delay. A delay adds phase lag that grows without limit as frequency rises, while leaving the gain untouched. It can erase a comfortable phase margin entirely, and it never shows up if you only look at a gain plot.

Try the interactive NovaSolver calculator

Sweeping a curve through the -1 point in your head is hard; watching it move is not. The Nyquist Diagram & Stability Margin Calculator on NovaSolver lets you choose an open-loop transfer function and adjust the gain K while the Nyquist plot redraws in real time, and it computes the gain margin, phase margin, and crossover frequencies automatically — with a stability verdict and a Bode plot comparison — so you can see exactly how raising the gain pushes the curve toward the critical point.

Related calculators

Bode plot generator — read the same margins from stacked gain and phase curves, which is often the easier view for loop shaping.
Root locus plotter — track where the closed-loop poles travel as gain rises, and see instability as a pole crossing into the right half-plane.
PID tuning calculator — once the margins tell you what the loop needs, tune the controller that delivers it.

You can browse the full set in the controls and frequency response tools hub.

Closing note

The Nyquist criterion is one of the most quietly powerful ideas in control engineering: it converts a closed-loop stability question into a geometry problem about one curve and one point. Stay clear of -1 and the loop is stable; the distance you keep is your margin. Aim for a phase margin near 45 degrees and a gain margin of 6 to 10 dB, read both numbers rather than one, and respect time delay as the silent margin-eater. Get the open-loop curve right and you will know how the closed loop behaves long before you commit to closing it.

Bode Plots: Reading How a System Responds, One Frequency at a Time

NovaSolver — Sun, 19 Jul 2026 12:17:03 +0000

Tap a wine glass and it rings at one clear pitch. Push a child on a swing and you quickly learn there is one rhythm that works and a dozen that fight you. Every physical system — an amplifier, a motor, a suspension, a thermostat — has a personality that depends entirely on how fast you ask it to do something. Push slowly and it follows faithfully. Push quickly and it lags, shrinks the response, or both. The Bode plot is the tool that draws that personality as a picture.

This article explains what a Bode plot actually shows, how to sketch one from a transfer function, and why the corner frequency of a first-order system is the single most useful landmark on the chart.

Why this calculation matters

Control engineers live in the frequency domain because it turns a hard question — "is this feedback loop stable, and how well damped is it?" — into something you can read off two curves. A Bode plot is two graphs stacked together: magnitude (in decibels) versus frequency, and phase (in degrees) versus frequency, both on a logarithmic frequency axis.

That format matters for a practical reason. In the frequency domain, cascaded blocks simply add. Multiply two transfer functions and their magnitude curves add in decibels while their phase curves add in degrees. A messy product of polynomials becomes a sum of simple shapes you can sketch by hand. From those shapes you read gain margin, phase margin, and crossover frequencies — the numbers that tell you whether closing the loop will give a crisp response, a sluggish one, or an oscillation that never settles.

The core method

Most systems can be built from a few standard pieces, and the most important is the first-order lag:

G(s) = 1 / (1 + s*tau)

Here tau is the time constant and s is the Laplace variable. To get the frequency response, substitute s = j*omega, where omega is angular frequency in rad/s. The magnitude and phase become:

|G| = 1 / sqrt(1 + (omega*tau)^2)
phase(G) = -arctan(omega*tau)

The behaviour splits cleanly into two regimes around one landmark — the corner frequency, also called the break frequency:

omega_c = 1 / tau          (rad/s)
f_c = 1 / (2*pi*tau)       (Hz)

Well below the corner, omega*tau is small, the magnitude is flat near its low-frequency value, and the phase is near zero — the system follows the input. Well above the corner, omega*tau dominates, the magnitude falls along a straight line of -20 dB per decade, and the phase settles toward -90 degrees — the system can no longer keep up.

Right at the corner, omega*tau equals 1. The magnitude is 1/sqrt(2), which is exactly -3 dB below the low-frequency gain, and the phase is precisely -45 degrees. That -3 dB, -45 degree point is the signature of every first-order lag, and recognizing it lets you read a time constant straight off a measured plot.

A worked example

Take a first-order system with the transfer function G(s) = 1/(1 + s*tau) and a time constant tau = 0.01 s. This is a realistic figure — it could be a small RC filter or the electrical response of a motor winding.

Step 1 — find the corner frequency. Convert the time constant to a frequency in hertz:

f_c = 1 / (2*pi*tau)
f_c = 1 / (2*pi*0.01)
f_c = 1 / 0.0628
f_c = 15.9 Hz

Step 2 — describe the response at the corner. At f_c = 15.9 Hz the magnitude has dropped to 3 dB below the low-frequency gain, and the phase lag has reached 45 degrees. This is the half-power point — the output carries half the power it would at very low frequency.

Step 3 — describe the response above the corner. Beyond 15.9 Hz the magnitude rolls off at a steady -20 dB per decade, meaning each tenfold increase in frequency divides the amplitude by ten. The phase continues past -45 degrees and approaches -90 degrees asymptotically.

So a single number, tau = 0.01 s, fixes the entire shape. The plot is flat to about 15.9 Hz, then tips into a -20 dB/decade slope, while the phase swings smoothly from 0 through -45 degrees at the corner toward -90 degrees. If you ever measure a flat region followed by a -20 dB/decade slope, you have found a first-order lag, and the corner tells you its time constant.

Common mistakes

Confusing rad/s and Hz. The corner is 1/tau in rad/s but 1/(2*pi*tau) in hertz. Forgetting the 2*pi factor shifts the corner by more than a factor of six and quietly invalidates everything downstream.

Reading the asymptotes as the exact curve. The straight-line approximation is a sketch, not the truth. The real magnitude is 3 dB below the asymptote intersection at the corner, and the phase is a smooth S-curve that has already moved before the corner and is still moving a decade after it.

Forgetting that phase leads magnitude into trouble. A first-order lag never reaches -180 degrees, so on its own it cannot make a loop unstable. Stability problems come from accumulated phase across several lags and any pure time delay. Watch the total phase, not one block's.

Using a linear frequency axis. Bode plots only show their characteristic straight-line slopes on a logarithmic frequency axis. Plotted linearly, the decade-by-decade structure collapses and the -20 dB/decade rule no longer looks like a line.

Treating decibels as a voltage ratio of 10. A decade is a factor of ten in frequency; -20 dB is a factor of ten in amplitude. But 20 dB, not 10 dB, corresponds to a tenfold amplitude change. Mixing the two scrambles every slope on the plot.

Try the interactive NovaSolver calculator

Sketching asymptotes by hand builds intuition, but seeing the exact curves move as you drag a parameter builds it faster. The Bode Plot Generator on NovaSolver plots gain and phase in real time for standard transfer functions — a first-order lag, a second-order system with adjustable damping, an integrator, a PD compensator, or a custom numerator and denominator — and it computes the gain margin, phase margin, and crossover frequencies automatically, with a stability assessment, so you can watch how the corner frequency and the rolloff move when you change a time constant or a gain.

Related calculators

Nyquist diagram & stability margin calculator — see the same open-loop response as a single polar curve and judge stability by how it encircles the critical point.
Bode lead-lag compensator designer — once you can read a Bode plot, use it to shape one: add phase where the loop needs it to recover margin.
Root locus plotter — the complementary view, tracking closed-loop pole positions as gain varies.

You can browse the full set in the controls and frequency response tools hub.

Closing note

A Bode plot is less a calculation than a way of seeing. It takes a transfer function — an abstract ratio of polynomials — and turns it into two honest curves that tell you what the system does at every speed. Learn to spot the first-order corner with its -3 dB, -45 degree fingerprint, learn that cascaded blocks just add, and learn that phase is what eventually bites you, and most loop-shaping work becomes a matter of reading the picture and nudging it. Build the asymptotes first, check them against the exact curves, and let the plot tell you where the margins are.

Retaining Wall Design: Checking a Wall Against Overturning, Sliding, and Bearing

NovaSolver — Sun, 19 Jul 2026 00:17:02 +0000

Drive past any hillside road cut and you will see them: concrete walls holding back a slope that would otherwise slump onto the pavement. Most people never give them a second look. But behind each wall is a quiet, continuous push from the soil — a force that never lets up, never goes on holiday, and grows with the square of the wall height. A retaining wall does not fail because of one dramatic event. It fails because that steady push slowly wins.

This article walks through how a cantilever or gravity retaining wall is checked for stability. We will turn backfill into a measurable thrust, set up the three classic failure modes, and work a full numerical example you can reproduce by hand.

Why this calculation matters

A retaining wall has only one job: to resist lateral earth pressure without moving in a way that endangers what sits behind or in front of it. There is no redundancy. If the wall rotates, slides, or sinks unevenly, the consequences are immediate and visible — cracked pavements, tilted fences, and in the worst cases a full collapse that brings the retained slope down with it.

The reason the calculation deserves care is that earth pressure is not intuitive. The thrust on a 5 m wall is not 25 percent larger than the thrust on a 4 m wall; it is roughly 56 percent larger, because pressure grows linearly with depth and the total force integrates that triangle. Designers who scale a wall by eye tend to be optimistic. A structured stability check — overturning, sliding, bearing — replaces that guesswork with three numbers you can defend in a design review.

The core method

Start with the lateral pressure the backfill exerts. For granular soil with friction angle phi, level ground, and a smooth vertical wall, Rankine's active earth pressure coefficient is:

Ka = (1 - sin(phi)) / (1 + sin(phi))

Equivalently, Ka = tan^2(45 - phi/2). The active pressure at depth z is Ka times gamma times z, where gamma is the soil unit weight. Integrating that triangular distribution over the wall height H gives the total active thrust per metre of wall:

Pa = 0.5 * Ka * gamma * H^2

Because the pressure diagram is a triangle, its resultant acts at one third of the height above the base — at H/3. That lever arm is what turns the thrust into an overturning moment about the toe (the front bottom edge of the wall):

Mo = Pa * (H/3)

Resisting that rotation is the weight of the wall and its base footing. The total weight W acts down through the combined centre of gravity, located some horizontal distance from the toe. That gives the resisting moment:

Mr = W * (lever arm of W from the toe)

The factor of safety against overturning is simply the ratio:

FS_overturning = Mr / Mo

Two more checks complete the picture. Sliding is resisted by friction along the base — the base friction force divided by the horizontal thrust gives FS against sliding, often required to be at least 1.5. Bearing capacity is checked by comparing the peak contact pressure under the toe against the allowable bearing pressure of the foundation soil. A wall must pass all three; a comfortable overturning margin does not excuse a weak base.

A worked example

Consider a cantilever retaining wall of height H = 5 m holding back granular backfill. The soil has a friction angle phi = 30 degrees and a unit weight gamma = 18 kN/m^3.

Step 1 — earth pressure coefficient. For phi = 30 degrees:

Ka = (1 - sin 30) / (1 + sin 30) = (1 - 0.5) / (1 + 0.5) = 0.333

Step 2 — active thrust. Integrate the pressure triangle over the 5 m height:

Pa = 0.5 * Ka * gamma * H^2
Pa = 0.5 * 0.333 * 18 * 25 = 74.9 kN/m

That is the horizontal push per metre run of wall. It acts at H/3 = 1.67 m above the base.

Step 3 — overturning moment about the toe.

Mo = Pa * (H/3) = 74.9 * 1.67 = 124.8 kN.m per metre

Step 4 — resisting moment. The wall stem and base footing together weigh W = 180 kN/m, and their combined centre of gravity sits 1.6 m horizontally from the toe:

Mr = W * 1.6 = 180 * 1.6 = 288 kN.m per metre

Step 5 — factor of safety against overturning.

FS = Mr / Mo = 288 / 124.8 = 2.3

A factor of 2.3 clears the usual minimum of 2.0, so this wall is safe against rotation about its toe. The next steps would repeat the logic for sliding and bearing, but the overturning result already shows the structure of the method: every check is a ratio of what resists to what drives.

Common mistakes

Forgetting surcharge loads. A road, a stockpile, or a parked vehicle on the backfill adds a uniform pressure that does not vanish near the surface. Surcharge raises the resultant and shifts it upward, increasing the overturning moment more than its magnitude alone suggests.

Using active pressure where it cannot develop. Rankine active pressure assumes the wall yields slightly outward, letting the soil relax to its minimum state. A wall braced at the top, or one that genuinely cannot move, sees at-rest pressure instead — noticeably higher. Choosing the wrong coefficient understates the load.

Ignoring water behind the wall. Drainage is structural, not cosmetic. If the backfill saturates, hydrostatic pressure adds to the soil thrust and the effective unit weight changes. A blocked weep hole can quietly double the design load.

Checking overturning and stopping there. The three failure modes are independent. A wall can be generously stable against rotation yet slide on a slick clay foundation, or punch into soft ground. All three factors of safety must be satisfied.

Treating the friction angle as exact. Phi is measured with scatter and can fall if the soil is disturbed or wetted. A small drop in phi raises Ka and the thrust together, so a thin margin on paper can disappear in the field.

Try the interactive NovaSolver calculator

Running the moment balance once by hand is the right way to learn it, but design is iterative — you change the base width, adjust the stem, and want all three checks to update at once. The Retaining Wall Design Calculator on NovaSolver does exactly that: pick a wall type and either Rankine or Coulomb earth pressure theory, set the geometry, soil properties, surcharge, and allowable bearing pressure, and it returns the earth pressure coefficient, the active force, and live factors of safety for overturning, sliding, and bearing — with a Mononobe-Okabe seismic option when you need to add a horizontal acceleration coefficient.

Related calculators

Earth pressure calculator — focus purely on the lateral pressure side: compare active, at-rest, and passive states for different soils and wall conditions.
Bearing capacity calculator — size the foundation that carries the wall, so the bearing check has a defensible allowable pressure behind it.
Slope stability calculator — when the wall sits in or below a slope, check the global failure surface that can pass beneath the whole structure.

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

Closing note

Retaining wall design is one of those topics that looks heavy but rests on a handful of clear ideas. Soil pushes with a thrust that grows as the square of height. That thrust acts at one third of the height and tries to rotate the wall about its toe. Weight resists. Friction resists sliding. The footing resists bearing failure. Each check is a ratio, and a wall is only as safe as its weakest ratio. Compute all three, give yourself an honest margin on the friction angle, keep the drainage working, and the wall will go on doing its quiet job for decades.

Lateral Earth Pressure: How Hard Does Soil Push on a Wall?

NovaSolver — Sat, 18 Jul 2026 12:17:03 +0000

Stand a vertical wall against a bank of soil and the soil does not just sit there. It leans on the wall, and it leans hard. That sideways push is what topples basement walls during construction, bows out retaining structures over the years, and blows out unbraced excavations without warning. Soil behaves partly like a fluid — the deeper you go, the harder it pushes — but unlike water, the pressure also depends on the soil's own internal friction and on which way the wall is moving.

This article explains lateral earth pressure: where it comes from, how Rankine's theory turns soil properties into a pressure you can design for, a full worked example on a granular backfill, and the mistakes that most often lead to an underdesigned wall.

Why this calculation matters

Lateral earth pressure is the load case behind a long list of structures: retaining walls, basement and foundation walls, bridge abutments, sheet-pile cofferdams, and braced excavations. Every one of them holds back a body of soil, and every one of them must be sized for the horizontal thrust that soil delivers. Underestimate the pressure and the wall slides, overturns, or cracks. Overestimate it and the wall becomes needlessly massive and expensive.

What makes earth pressure more subtle than a simple fluid load is that its magnitude depends on movement. A wall that yields slightly away from the soil lets the backfill relax into its active state, the lowest pressure it can exert. A wall pushed into the soil mobilises the passive state, the highest resistance the soil can offer. A wall held perfectly rigid sees the at-rest pressure, somewhere in between. Knowing which state applies — and designing for it — is central to getting a retaining structure right.

The core formula

Rankine's theory, published in 1857, gives a clean way to estimate earth pressure for the idealised case of a smooth vertical wall retaining horizontal backfill. The pressure at any depth is the vertical stress multiplied by an earth pressure coefficient that depends on the soil friction angle phi.

For the active case — the wall yielding away from the soil — the coefficient is:

Ka = (1 - sin(phi)) / (1 + sin(phi))

The passive coefficient, for a wall pushed into the soil, is the reciprocal:

Kp = (1 + sin(phi)) / (1 - sin(phi))

The contrast is large. A soil with a friction angle of 30 degrees gives Ka = 0.333 and Kp = 3.0 — a ninefold difference. That is why passive resistance in front of a wall toe can be a major stabilising force, and why losing it, through erosion or excavation, is so dangerous.

For a dry granular backfill with no surcharge, the active pressure increases linearly with depth z:

p = Ka*gamma*z

where gamma is the soil unit weight. The pressure forms a triangular distribution, zero at the top and largest at the base, p = Ka*gamma*H. Because the distribution is triangular, the total active thrust per unit length of wall is the area of that triangle:

Pa = 0.5*Ka*gamma*H^2

and it acts at the centroid of the triangle, one third of the wall height above the base.

A worked example

Take a smooth vertical wall retaining a dry granular backfill. The soil has a friction angle phi = 30 degrees and a unit weight gamma = 18 kN/m^3, and the wall retains the soil to a height H = 4 m.

Step 1 — compute the active earth pressure coefficient.

Ka = (1 - sin(30 degrees)) / (1 + sin(30 degrees))
Ka = (1 - 0.5) / (1 + 0.5)
Ka = 0.5 / 1.5 = 0.333

Step 2 — find the active pressure at the base. The pressure grows linearly with depth and reaches its maximum at z = H:

p = Ka*gamma*H = 0.333 * 18 * 4 = 24 kPa

Step 3 — compute the total active thrust per metre of wall. This is the area of the triangular pressure diagram:

Pa = 0.5*Ka*gamma*H^2
Pa = 0.5 * 0.333 * 18 * 16
Pa = 48 kN/m

Step 4 — locate the thrust. The resultant acts at the centroid of the triangle, one third of the height above the base:

H/3 = 4 / 3 = 1.33 m above the base

So this wall must resist a horizontal thrust of 48 kN per metre of length, applied 1.33 m above its base. That lever arm matters as much as the force itself: the overturning moment about the wall toe is the product of the two, and it is the moment, not the force alone, that decides whether the wall tips.

Common mistakes

Designing for the wrong pressure state. Using the at-rest coefficient where the wall can yield wastes material; using the active coefficient on a rigid, unyielding wall underestimates the load. The wall's restraint and expected movement determine which coefficient applies.

Placing the thrust at mid-height. Because the active pressure distribution is triangular, not rectangular, its resultant acts at one third of the height from the base — not at the middle. Putting it at mid-height understates the overturning moment.

Ignoring water behind the wall. Rankine's dry-backfill formula leaves out pore water entirely. If drainage fails and the backfill saturates, full hydrostatic water pressure adds to the soil pressure, and water pushes with a coefficient of 1.0, far higher than Ka. Poor drainage is one of the most common causes of retaining wall failure.

Forgetting surcharge loads. Traffic, stockpiled material, or a building near the top of the wall adds a vertical surcharge that raises the lateral pressure at every depth. A surcharge q adds a roughly uniform pressure of Ka*q across the full wall height.

Counting on passive resistance that may not be there. Passive pressure in front of the wall toe is a genuine stabilising force, but only while the soil that provides it stays in place. Future excavation, scour, or erosion can remove it, so it is often discounted or used cautiously.

Try the interactive NovaSolver calculator

Hand calculation is the right way to learn the method, but comparing cases — different friction angles, wall heights, and surcharges — is far faster interactively. The Rankine Earth Pressure Simulator on NovaSolver computes lateral earth pressure for a vertical wall with horizontal backfill — set the soil friction angle, unit weight, wall height, and surcharge, and it returns the active and passive coefficients Ka and Kp, the active and passive forces Pa and Pp, and a live diagram of the vertical pressure distribution along with the coefficient curves against friction angle.

Related calculators

Retaining wall calculator — turns the earth thrust into checks on sliding, overturning, and bearing for a complete wall design.
Slope stability analysis — for deciding whether soil can stand at its natural angle before a wall is even needed.
Bearing capacity calculator — to confirm the soil beneath the wall base can carry the vertical load the wall delivers.

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

Closing note

Lateral earth pressure is the load that defines retaining structures, and Rankine's theory gives an honest first estimate of it from just two soil properties — friction angle and unit weight — plus the wall height. The key ideas are worth carrying with you: pressure grows linearly with depth, the active state is far gentler than the passive state, the resultant thrust sits one third of the way up the wall, and water behind a wall changes everything. Compute the thrust, find its line of action, account for drainage and surcharge, and the retaining wall design that follows rests on solid ground.

Slope Stability and the Factor of Safety: When Will a Slope Slide?

NovaSolver — Sat, 18 Jul 2026 00:17:01 +0000

After a long spell of rain, a section of hillside above a road lets go. A wedge of soil slides downslope along a curved surface, burying the carriageway and shearing through anything in its path. Landslides like this are not random acts of nature — they happen when the forces driving a soil mass downhill finally exceed the forces resisting it. Slope stability analysis is the engineering discipline that predicts that tipping point before it arrives.

This article explains what the factor of safety against sliding actually measures, how to compute it for the simplest and most instructive case, works a full example, and lists the errors that most often make a slope look safer on paper than it is in the ground.

Why this calculation matters

Slopes are everywhere in civil engineering: highway and railway cuttings, earth dams, embankments, levees, mine spoil heaps, and the natural hillsides that infrastructure has to cross. Every one of them is a soil mass held in place only by its own shear strength. When that strength is exceeded along some surface, the mass moves — sometimes slowly over months, sometimes catastrophically in seconds.

The factor of safety is the single number that quantifies how close a slope is to that moment. It frames decisions about how steep a cut can be, how an embankment should be built, whether a slope needs drainage or a retaining structure, and how much warning a marginal hillside is giving. Because the consequences of a slope failure range from costly to fatal, the factor of safety is also a communication tool: it lets engineers, regulators, and clients agree on how much margin is enough.

The core method

Slope stability is fundamentally a comparison of two quantities along a potential failure surface:

FS = (resisting forces) / (driving forces)

The driving force is the component of the soil weight that acts down the slope. The resisting force is the shear strength of the soil mobilised along the failure surface. When FS is greater than 1, resistance wins and the slope holds. When FS falls to 1, the two are balanced and failure is imminent. Below 1, the slope is already moving.

The cleanest case to study is the infinite slope: a long, planar slope where the failure surface runs parallel to the ground at constant depth, so conditions repeat from one slice to the next. For a dry, cohesionless soil — clean sand or gravel with no cohesion and no pore water — the analysis collapses to a strikingly simple result:

FS = tan(phi) / tan(beta)

Here phi is the soil friction angle, the property that sets its shear strength, and beta is the slope angle measured from the horizontal. The weight of the soil and the depth of the surface both cancel out, leaving a pure comparison of two angles.

The physical reading is direct. A dry granular slope is stable as long as it is flatter than the soil's friction angle. The steepest stable slope is the one where beta equals phi — that limiting angle is the angle of repose, the same angle you see when sand is poured into a cone. Real slopes with cohesion, layering, or groundwater need fuller methods, such as the method of slices, but the infinite-slope formula captures the essential balance and is the right place to build intuition.

A worked example

Consider a long, planar slope cut into a dry cohesionless soil. The soil friction angle is phi = 32 degrees, and the slope is inclined at beta = 20 degrees to the horizontal.

Step 1 — confirm the case. The soil is dry and cohesionless, the slope is long and planar, so the infinite-slope formula applies directly.

Step 2 — evaluate the two tangents.

tan(phi) = tan(32 degrees) = 0.6249
tan(beta) = tan(20 degrees) = 0.3640

Step 3 — form the factor of safety.

FS = tan(phi) / tan(beta)
FS = 0.6249 / 0.3640
FS = 1.72

Step 4 — interpret the result. A factor of safety of 1.72 means the available shear resistance is about 72 percent greater than the force driving the slope downhill. The slope is stable. Had the result come out near 1.0, failure would be imminent and the slope would need to be flattened, drained, or supported. Had it fallen below 1.0, the slope could not stand at that angle at all.

It is worth noticing how sensitive the answer is to the slope angle. Steepen this slope from 20 to 30 degrees and tan(beta) rises to 0.577, dropping FS to about 1.08 — barely stable. A modest change in geometry moves a comfortable slope to the edge.

Common mistakes

Treating any FS above 1 as safe enough. A factor of safety of 1.05 is technically stable but offers almost no margin against the uncertainty in soil strength, geometry, and loading. Design practice typically calls for FS in the range of about 1.3 to 1.5 for long-term slopes, higher where the consequences of failure are severe.

Ignoring pore water pressure. The simple formula here assumes a dry slope. Rain and rising groundwater raise pore pressure, which cuts the effective stress and therefore the available shear strength. Many slopes that stand all summer fail after prolonged wet weather precisely because water, not geometry, changed.

Using a single failure surface. A slope does not get to choose a convenient surface — it fails along the most critical one. A proper analysis searches many candidate surfaces and reports the lowest factor of safety, not the first one tried.

Applying the infinite-slope formula where it does not belong. That formula is for long, planar, shallow failures in uniform soil. A deep rotational slip in clay, or a layered slope with a weak seam, needs the method of slices or a numerical analysis. Using the wrong model gives a confident but meaningless number.

Assuming soil strength is constant over time. Friction angle and cohesion can degrade through weathering, softening, or progressive failure. A slope that was safe when built is not guaranteed to stay safe decades later.

Try the interactive NovaSolver calculator

The infinite-slope formula builds intuition, but real slopes with curved failure surfaces and groundwater need a fuller treatment. The Slope Stability Analysis (Bishop Method) on NovaSolver does this interactively — set the slope height and angle, the soil cohesion, friction angle and unit weight, and the groundwater level, choose the number of slices, and it computes the factor of safety for a circular slip surface and sweeps for the minimum FS, reporting the critical circle's centre and radius alongside a visualisation of the slip surface.

Related calculators

Bearing capacity calculator — for a footing placed near or on a slope, where bearing failure and slope failure can interact.
Earth pressure simulator — to quantify the lateral pressures that develop when soil is retained rather than left to stand at its natural angle.
Retaining wall calculator — when a slope is too steep to stand alone and needs a structural wall to hold it back.

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

Closing note

Slope stability comes down to a contest between weight pulling soil downhill and shear strength holding it in place, and the factor of safety is the scoreboard. The infinite-slope result — FS equals tan(phi) over tan(beta) — is as compact as engineering formulas get, and it carries a memorable lesson: a dry granular slope is stable only while it is flatter than the soil's friction angle. Real slopes add cohesion, water, and curved failure surfaces, but the underlying question never changes. Compute the factor of safety, leave a margin that matches the consequences, and respect what water can do to a slope that looked fine yesterday.