Does a Bigger Turbo Really Mean More Lag? We Sized Five and Timed Them
Five turbo wheels, 60 to 110 mm, on one 3.0 L six at the same 1.92 bar target. We timed each to boost at four speeds and found where lag turns into never.
Ask a forum whether a bigger turbo means more lag and you get a yes, a story about a friend's build, and no numbers. Nobody in the thread has timed it, because timing it on a real engine means five turbochargers, five manifold fabrications and a dyno day for each.
In our simulator the compressor wheel is a slider. Load the turbocharged 3.0 L inline-six, hold it at 3,500 rpm on the dyno and snap the throttle open. With the stock 82 mm wheel the manifold crosses half a bar of boost at 0.33 s. Drag the slider to 60 mm and the same crossing comes at 0.10 s. Drag it to 110 mm and you can wait the full eight seconds of the test: the manifold settles at +0.22 bar and never reaches half a bar. On this engine, lag is a delay at the small end of the slider and a refusal at the big end.
Key Takeaways
- Bigger means more lag at every speed we tested, and the penalty outruns the size change. At 3,500 rpm, time to +0.5 bar of boost runs 0.10 / 0.17 / 0.33 / 0.82 s for 60, 70, 82 and 95 mm wheels, and the 110 mm wheel never gets there. Time to the 1.92 bar target: 0.12 / 0.21 / 0.42 / 1.36 s, then never.
- Rotor inertia rises ×20.7 from 60 to 110 mm, following the fifth power of diameter. Turbine throat area rises ×3.4, the square. The slowest wheel is also the freest-breathing one: 2.02 bar of exhaust backpressure at 7,000 rpm against 5.14 for the 60 mm.
- Below 5,000 rpm the 110 mm wheel never holds half a bar, and it holds the 1.92 bar target only from 5,000 rpm up. Torque at 2,500 rpm: 512.1 N·m on the 60 mm wheel, 247.0 on the 110.
- The 70 mm wheel reads the most peak power, 429.63 hp against 385.48 stock, because its wastegate is sized off the wheel, runs out of travel and lets the manifold creep to 2.61 bar. The simulator charges for the backpressure that comes with it, which is why the 60 mm wheel, creeping to 2.78 bar with up to 5.14 bar of exhaust behind it, peaks at 5,060 rpm and falls behind at the top. It does not charge for heat. Treat the small wheels as the fast end of the lag curve.
- Two 59.5 mm wheels spool slower than one 75.0 mm wheel in this model, 1.15 s against 0.86 s to 90% at 2,500 rpm, for the same power to about 1%. The real twin-turbo advantage lives in manifold packaging the model does not have.
On this page: Quick answer · Why inertia follows the fifth power · Five wheels on one engine · The throttle step · Held wide open · One wheel or two · The folklore · What we do not model · Try it in the simulator · FAQ · How these numbers were made
Quick Answer: Does a Bigger Turbo Mean More Lag
Yes, at every engine speed we tested, and the penalty grows faster than the wheel does. On the same 3.0 L six at the same 1.92 bar target, a 60 mm compressor wheel reaches +0.5 bar of boost in 0.10 s at 3,500 rpm, the stock 82 mm wheel in 0.33 s, a 95 mm wheel in 0.82 s, and a 110 mm wheel never does inside our eight-second window. Rotor inertia between the smallest and largest wheel differs by ×20.7, and lag follows it.
The delay is the smaller half of the story. At low engine speed the big wheel never arrives. The 110 mm wheel never holds half a bar of boost below 5,000 rpm and holds the 1.92 bar target only from 5,000 up. At 2,500 rpm the stock wheel holds +0.57 bar, the 95 mm +0.15 and the 110 mm +0.06: an 8.5:1 engine with a restriction in the exhaust. Every figure in this article is this model's, measured on 2026-10-07, and the section on what the simulator does not model marks where the small wheels' numbers stop meaning anything.
Why Inertia Grows With the Fifth Power of Diameter
A turbocharger's compressor and turbine share one shaft, so the compressor cannot make pressure until exhaust energy has spun the whole rotor up (Garrett Motion, What is a turbo and how does it work?, retrieved 2026-09-11). Turbo lag is the time between opening the throttle and boost arriving, and it exists because the rotor has inertia and the exhaust flow available to accelerate it from low load is small (Wikipedia, Turbocharger, retrieved 2026-09-11). How a turbocharger works shows one measured trace of that acceleration on the stock wheel and covers the wastegate and intercooler. This article changes the wheel and leaves the rest alone.
Inertia scales harder with size than the eye expects. The moment of inertia of a solid disc is half its mass times the square of its radius (Wikipedia, List of moments of inertia, retrieved 2026-09-11). Scale a wheel up with its proportions fixed and its mass grows with the cube of the diameter, so mass times radius squared grows with the fifth power. From 60 to 110 mm in our simulator that is ×20.7. The turbine throat, which sets how much exhaust the wheel swallows, grows with the square of the diameter, ×3.4 over the same range. The big wheel gets 3.4 times the exhaust window and 20.7 times the flywheel.
Our model builds the whole turbo from the compressor diameter by that rule: the turbine wheel at 0.88× the compressor, rotor inertia proportional to D^5, the turbine throat and the wastegate port each a fixed fraction of the turbine wheel's disc, and a 550 m/s tip-speed limit that gives each wheel its own top rpm. Real wheels are not solid discs, so their exponent sits below five. The fifth power is the model's. The wastegate port shrinking with the wheel matters later.
Five Wheels on One Engine
turbo_i6_3L is our turbocharged 3.0 L inline-six: 86 × 86 mm, 8.5:1, redline 7,000 rpm, target 1.92 bar absolute, intercooler effectiveness 0.78. We built it five times with the compressor wheel at 60, 70, 82 (stock), 95 and 110 mm and changed nothing else. Everything below derives from that one number.
| Wheel | Turbine | Inertia (kg·m²) | Throat (mm²) | Tip limit (rpm) | Peak hp @ rpm | Peak N·m @ rpm |
|---|---|---|---|---|---|---|
| 60 mm | 52.8 mm | 6.7e-6 | 375 | 175,070 | 407.91 @ 5,060 | 639.38 @ 3,983 |
| 70 mm | 61.6 mm | 14.6e-6 | 511 | 150,060 | 429.63 @ 7,000 | 568.59 @ 4,845 |
| 82 mm (stock) | 72.2 mm | 32.2e-6 | 701 | 128,100 | 385.48 @ 7,000 | 541.98 @ 4,629 |
| 95 mm | 83.6 mm | 67.1e-6 | 941 | 110,571 | 369.56 @ 7,000 | 517.20 @ 4,629 |
| 110 mm | 96.8 mm | 139.7e-6 | 1,261 | 95,493 | 351.42 @ 7,000 | 489.18 @ 4,629 |
Start with the two right-hand columns. Peak torque falls as the wheel grows, from 639.38 N·m to 489.18. Peak power does not: the 70 mm wheel tops it at 429.63 hp, and the 60 mm wheel makes its 407.91 at 5,060 rpm and falls away after. The two wheels a forum would call "too small" both beat the stock wheel, and that result is a warning with its own section under the held-throttle results.
For scale, the same six with the turbo removed, at the same 8.5:1, makes 162.70 hp at 4,845 rpm and 246.88 N·m at 4,198. The stock wheel more than doubles it. How much horsepower one point of compression adds measures why a boosted engine sits at 8.5:1 in the first place.
The Throttle Step: Seconds to Boost
The test is the one in the turbocharger post, run twenty times. Hold the engine at a fixed rpm with the throttle shut for three seconds, then snap it wide open and record manifold pressure for eight. Before the step the manifold sits at 0.087 bar absolute on every wheel, a closed-throttle vacuum. The rotor before the step at 2,500 rpm is idling at 9,226 rpm on the 60 mm wheel, 2,016 on the stock wheel and 403 on the 110.
We time two fixed thresholds: the first crossing of +0.5 bar of boost and the first crossing of the 1.92 bar target, which is +0.92 bar. Half a bar is just over half the way there. "Never" means the manifold did not get there inside the eight seconds.
| Wheel | 2,000 rpm | 2,500 rpm | 3,500 rpm | 5,000 rpm |
|---|---|---|---|---|
| 60 mm | 0.39 / 0.51 | 0.21 / 0.27 | 0.10 / 0.12 | 0.05 / 0.07 |
| 70 mm | 1.16 / 3.37 | 0.41 / 0.54 | 0.17 / 0.21 | 0.09 / 0.11 |
| 82 mm (stock) | never / never | 1.84 / never | 0.33 / 0.42 | 0.15 / 0.20 |
| 95 mm | never / never | never / never | 0.82 / 1.36 | 0.30 / 0.38 |
| 110 mm | never / never | never / never | never / never | 0.73 / 1.16 |
Seconds to +0.5 bar of boost / seconds to the 1.92 bar target after a throttle step at a held engine speed.
Read the 3,500 rpm column down. To half a bar: 0.10, 0.17, 0.33, 0.82 s, never. To the target: 0.12, 0.21, 0.42, 1.36 s, never. The step from the stock wheel to the 95 mm more than doubles the wait to half a bar and more than triples the wait to the target; the step to 110 removes both. Read the rows across and each wheel has a speed below which a threshold slips away: the 70 mm wheel needs 3.37 s to reach the target at 2,000 rpm, the stock wheel never reaches it at 2,500 and reaches neither threshold at 2,000, the 95 mm reaches neither at 2,500, and the 110 mm reaches the target only at 5,000. The final boost the manifold settles at tells you why:
| Wheel | 2,000 rpm | 2,500 rpm | 3,500 rpm | 5,000 rpm |
|---|---|---|---|---|
| 60 mm | +1.047 | +1.170 | +1.560 | +1.757 |
| 70 mm | +0.925 | +1.031 | +1.204 | +1.435 |
| 82 mm (stock) | +0.163 | +0.570 | +1.061 | +1.188 |
| 95 mm | +0.068 | +0.152 | +0.943 | +1.059 |
| 110 mm | +0.030 | +0.063 | +0.222 | +0.944 |
Boost in bar, mean over the last second of the eight, after the step.
At 2,500 rpm the stock wheel settles at +0.570 bar and the 110 mm at +0.063. There is no lag to measure on the big wheel there, because there is no boost coming. A longer window would not help: the held-throttle table below shows the 110 mm wheel settling at 1.062 bar absolute at 2,500 rpm with no more to come. Read the 60 mm row the other way: +1.757 bar at 5,000 rpm is 0.84 bar past the target, and the held-throttle section explains where it comes from.
One more row from the traces. Each wheel refills the charge pipe to atmospheric before it makes boost, and each bigger wheel does that later: 0.07, 0.10, 0.14, 0.20 and 0.29 s at 2,500 rpm, where the naturally aspirated six's plenum is back at atmospheric within about 25 ms on the same throttle step. Part of what you feel as lag on a big turbo is the pipe between the compressor and the throttle refilling behind a wheel that had all but stopped.
Why 50% and 90% of Final Boost Mislead Across Wheels
The turbocharger post timed the stock wheel to half and to 90% of its final boost, which is the honest metric for one wheel at one speed. Across wheels it lies, because the final boost differs. The stock wheel at 2,500 rpm settles at +0.570 bar: its 50% point comes at 0.76 s, early, while its half-bar point comes at 1.84 s and its 90% point at 2.01 s. A wheel that reaches less boost hits "half of what it will do" sooner and "90% of what it will do" later, and neither number says whether the driver got any boost. The tables above use fixed thresholds for that reason.
We also re-ran every test in this article on the simulator build the article was first published from, on 2026-09-11, as a control. It reproduces every figure published then within one 4 ms sample, so every change since is the model's: a physical throttle and intake manifold, which refill faster from a closed throttle and leave the rotor spinning faster before the step; exhaust temperature that follows the burn, which drives a small turbine harder; and backpressure that now costs pumping work and leaves exhaust gas in the cylinder. Today the stock wheel at 2,500 rpm reads 0.14 s to atmospheric, 0.76 s to half its final boost and 2.01 s to 90%, with the rotor climbing from 2,016 to 73,180 rpm. The turbocharger post's trace was measured on an older build and reads differently: 0.30 s to atmospheric, 0.75 s to half and 1.55 s to 90% of a lower +0.38 bar.
Held Wide Open: What Each Wheel Gives Back
Lag is the transient. The steady state is what you get once the rotor has caught up, and it moves the same way. Each wheel held wide open at eight engine speeds for six seconds, torque and pressures averaged over the last three:
| Wheel | Holds 1.92 bar from | N·m at 2,500 | N·m at 4,000 | MAP at 7,000 | Backpressure at 7,000 | Wastegate at 7,000 | Rotor at 7,000, % of tip limit |
|---|---|---|---|---|---|---|---|
| 60 mm | 2,000 rpm | 512.1 | 640.3 | 2.727 bar | 5.14 bar | 100% | 105.9% |
| 70 mm | 2,000 rpm | 482.0 | 560.7 | 2.613 bar | 3.85 bar | 100% | 96.7% |
| 82 mm (stock) | 3,000 rpm | 371.8 | 536.3 | 2.232 bar | 2.73 bar | 80% | 83.4% |
| 95 mm | 4,000 rpm | 270.3 | 507.3 | 2.102 bar | 2.31 bar | 47% | 78.1% |
| 110 mm | 5,000 rpm | 247.0 | 362.5 | 1.983 bar | 2.02 bar | 17% | 73.8% |
The second column is the one to remember. The rpm at which each wheel first holds its 1.92 bar target walks up the rev range with diameter: 2,000, 2,000, 3,000, 4,000 and 5,000. That is the same fact as the "never" cells in the lag table, read from the other side. A wheel that cannot hold the target at a given rpm in steady state has nothing to spool up to at that rpm in a transient.
Torque at 2,500 rpm tells the driver's version. The 60 mm wheel makes 512.1 N·m there, the stock wheel 371.8, the 110 mm 247.0, which is within a newton-metre of what the naturally aspirated six makes at its own torque peak. At 1,500 rpm the 110 mm wheel reads 219.9 N·m with a manifold pressure of 1.012 bar: an unboosted 8.5:1 engine pushing against 1.07 bar of exhaust. By 4,000 rpm the 70, 82 and 95 mm wheels sit within 54 N·m of each other, 560.7 down to 507.3; the 60 mm wheel reads 640.3 on 2.74 bar of manifold pressure, and the 110 mm is at 362.5.
The right-hand columns are the bill. Exhaust backpressure at redline runs from 5.14 bar on the 60 mm wheel to 2.02 on the 110, and the pumping cost runs from +1.746 bar of mean effective pressure to +0.042. The 60 and 70 mm wastegates are 100% open at 7,000 rpm with nowhere left to go. The 110 mm gate has only just started to open, at 17%, because below 5,000 rpm the turbo never reaches the pressure it is set for.
The Small-Wheel Trap: 430 hp From Boost the Gate Cannot Hold
The 70 mm wheel makes 429.63 hp against 385.48 stock, and the 60 mm wheel 407.91, and neither number means what it seems to. In this model the wastegate port is sized off the turbine wheel, so a small wheel gets a small gate. The 70 mm gate is flat out from 5,000 rpm up and the manifold creeps past the 1.92 bar target to 2.44 to 2.61 bar. The 60 mm gate is flat out from 4,000 rpm up and the manifold sits at 2.73 to 2.78. That is up to +0.86 bar of boost the engine was never set for, and the extra power is that boost.
The simulator does charge for what comes with it. A turbine that small is a restriction: the 60 mm wheel holds 5.14 bar of exhaust backpressure at 7,000 rpm against 2.73 on the stock wheel. The pistons pay to push the exhaust out against it, +1.746 bar of mean effective pressure against +0.372, and the exhaust left in the cylinder above intake pressure takes room from the next charge. That bill is why the 60 mm wheel's power peaks at 5,060 rpm and its torque at redline is 404.8 N·m, behind the 70 mm's 436.6. On the 70 mm wheel the bill is smaller, 3.85 bar of backpressure, and the extra boost still wins.
What it does not charge is heat, and you should know that before you copy the setting. There is no exhaust-temperature limit, and the 550 m/s tip-speed limit is a warning in the model, not a failure: the 60 mm rotor runs past it from 5,000 rpm up and reaches 105.9% at redline. Compressor efficiency at redline has slid to 0.603 on the 60 mm wheel against 0.748 on the stock wheel. Settled knock read under 0.001 at every held point on every wheel on the preset's 98 octane.
On a real engine a wheel this far off its map would be into choke on the compressor side, where the map's high-flow edge caps what the wheel can pass (Wikipedia, Compressor map, retrieved 2026-09-11), cooking on the turbine side, and fitted with a wastegate sized for the engine rather than the wheel. The 60 and 70 mm runs show the fast end of the lag curve.
One Wheel or Two
The turbocharger post lists two small turbos among the ways manufacturers attack lag. We tested it on the same six the way the simulator fits turbos itself: no explicit wheel size, each wheel sized from the displacement it serves, housings flow-matched, total throat area conserved. One turbo on all six cylinders gets a 75.0 mm wheel. Two turbos on three cylinders each get 59.5 mm wheels.
| Layout | Wheel | Inertia each | Throat each | Peak hp | Peak N·m | 2,500 rpm: t50 / t90 / target | 3,500 rpm: t50 / t90 / target |
|---|---|---|---|---|---|---|---|
| Single | 75.0 mm | 20.6e-6 kg·m² | 586 mm² | 391.93 | 555.24 | 0.60 / 0.86 / 0.88 s | 0.23 / 0.28 / 0.28 s |
| Twin | 59.5 mm | 6.5e-6 kg·m² | 293 mm² | 396.25 | 553.33 | 0.75 / 1.15 / 1.20 s | 0.28 / 0.35 / 0.33 s |
Same power to about 1%, and the twin spools slower. At 2,500 rpm it reaches 90% of its final boost in 1.15 s against 0.86 for the single, and the 1.92 bar target in 1.20 s against 0.88. At 3,500 the gap narrows and the order holds.
The inertia and throat columns give the first-order account, and it comes out closer to a tie than the result. Each small wheel carries 0.315× the inertia of the single and has to reach 1.26× its shaft speed for the same pressure ratio, so the energy stored in each rotor at boost is 0.315 × 1.26², half the single's, and the pair store what the one does. The same exhaust supplies it, split two ways through two half-size throats. Whatever breaks the tie sits in the turbine and compressor maps at the smaller diameter, and we did not isolate it: the model's answer is the measured one, not a derived one.
That is the model's answer, and a bounded one. The simulator has one lumped exhaust manifold: no pulse separation, no manifold volume per bank, no runner length. The real twin-turbo advantage lives there. Two turbos sit closer to their three cylinders each, with shorter runners and less manifold volume to pressurise before the turbine sees anything, and a twin-scroll housing keeps one bank's exhaust pulse from cancelling the other's; a smaller upper-deck piping volume and twin-scroll housings both sit on the standard list of lag reductions (Wikipedia, Turbocharger, retrieved 2026-09-11). A mean-value model with one plenum cannot see any of that. On the mechanism it can see, two small wheels lose to one medium one.
What the Folklore Gets Right and Wrong
"A bigger turbo means more lag." Right, and an understatement. The forum version has lag as a delay that gets longer. In our tables it is a delay that gets longer until the threshold leaves the rev range: 0.33 s to half a bar at 3,500 rpm on the stock wheel, 0.82 s on the 95 mm, and no crossing at all on the 110 mm below 5,000 rpm. Sizing a turbo picks the engine speed below which you are driving a low-compression naturally aspirated engine.
"A small turbo runs out of breath up top." Right on a real engine, and our model now shows part of it: at 7,000 rpm the 60 mm wheel's wastegate is 100% open, its rotor is at 105.9% of its limit, its compressor efficiency has fallen to 0.603 and the exhaust is at 5.14 bar, and the pumping and residual-gas bill pulls its power peak back to 5,060 rpm. It still reads 407.91 hp, more than stock, because its undersized gate lets boost creep, and the model does not charge for heat.
"Two small turbos spool faster than one big one." In this model, no: 1.15 s against 0.86 s to 90% at 2,500 rpm for the same power. On a real car the answer is often yes, and the likeliest difference is the manifold the model does not have: shorter runners, less volume per turbine and pulse separation, none of which is a rotor-inertia effect.
"Modern turbos have no lag." Less than they had, and the reasons are the ones in the tables: smaller and lighter wheels, variable-geometry turbines and, on electrically assisted turbos, a motor on the shaft (Wikipedia, Turbocharger, retrieved 2026-09-11). None of them removes the rotor. Formula 1 makes the point in reverse: its 2026 power units dropped the MGU-H (Wikipedia, 2026 Formula One World Championship, retrieved 2026-09-11), the motor-generator on the turbocharger shaft between turbine and compressor (Wikipedia, Formula One engines, retrieved 2026-09-11), which could spin the compressor up before the exhaust did. Without it the compressor waits on exhaust energy like the one in this article, and turbo response is a place to win or lose time again. The Race's report on Mercedes' engine upgrade plan calls the turbo the most obvious area where good gains can be found (The Race, Mercedes' new engine upgrade plan is terrible news for Ferrari, retrieved 2026-09-11). That trade runs at the engine speeds covered in why F1 engines rev to 20,000 rpm.
A belt-driven compressor sidesteps the whole problem by taking its torque from the crank instead of the exhaust. That trade is the subject of the sister post published the same day as this one, Supercharger vs turbocharger: which makes more power?, and of the measured comparison in Roots vs centrifugal superchargers.
What This Simulator Does Not Model
The lag times above describe this model, with the compressor wheel as its one input. Six specific gaps:
- Rotor inertia follows D^5 by construction. That is the law a solid disc of fixed proportions obeys. Real wheels are not solid discs, so their inertia grows with a lower exponent, and the ×20.7 between 60 and 110 mm is the model's ratio. The ordering of the five wheels would survive a lower exponent; the spacing would not.
- The wastegate port is sized off the turbine wheel. A small wheel gets a small gate, runs it out of travel and creeps past the target, and that boost is most of the 60 and 70 mm wheels' extra power. A real gate is sized for the engine and housing it serves, and the simulator has no separate gate setting.
- Backpressure is charged, heat is not. A tight turbine costs pumping work and leaves exhaust gas in the cylinder, but there is no exhaust-temperature limit, no oil and no bearing losses that scale with speed. The 550 m/s tip-speed limit lights a warning rather than capping the rotor, and the 60 mm wheel runs 5.9% past it at redline.
- One lumped exhaust manifold. No pulse separation, no per-bank volume, no runner length, no twin-scroll. A real twin-turbo layout and a real twin-scroll housing win there, and the model cannot see it, so its twin result says nothing about them.
- Knock is calibrated at 4,400 rpm and above. Every held point read under 0.001 settled on 98 octane. Below 4,400 rpm and at part throttle the model is silent, so nothing here is a claim about knock on the launch.
- The thresholds are ours. +0.5 bar of boost, the 1.92 bar target and an eight-second window are the article's choices. A different pair of thresholds would move the numbers and keep the order.
The full scope, and the rest of what we get wrong, is in how we build the simulator. The compressor-side limits that a small wheel would hit in the real world, surge on the left of the map and choke on the right, are drawn in compressor stall vs surge.
Try It in the Simulator
Every run above reproduces in the browser, on the stock preset with one slider moved.
- Open the turbo six at stock, start it, and run the dyno once to see the 82 mm curve. Then find Compressor Ø in the spec panel; the slider runs from 40 to 110 mm. Hold the engine at 3,500 rpm on the dyno, snap the throttle open and watch the Boost row climb behind the Turbo row, which is rotor rpm. Boost should cross +0.5 bar at about 0.33 s and reach the 1.92 bar target at about 0.42 s.
- Load the 60 mm wheel and repeat the step. Half a bar arrives at 0.10 s. Then hold it at 7,000 rpm and read the Wastegate row at 100% and the Exhaust row against MAP: 5.14 bar against 2.727. The gate is flat out and the manifold has crept 0.8 bar past the 1.92 bar target.
- Load the 110 mm wheel and repeat the step at 3,500 rpm. Boost settles at +0.22 bar and stays there. Raise the hold to 5,000 rpm and watch half a bar arrive at 0.73 s. The Wastegate row reads 0% below 5,000 rpm, because down there the turbo never reaches the pressure the gate is set for.
- Load the twin 59.5 mm layout, hold 2,500 rpm and step the throttle. Boost reaches the 1.92 bar target at 1.20 s. Load one 75 mm wheel and the same crossing comes at 0.88 s.
The 1.92 bar target, the 550 m/s tip-speed limit, the D^5 inertia law, the 0.88× turbine-to-compressor ratio and the throat and wastegate fractions are simulator calibration; the slider range and the wastegate's proportional behaviour are the product's. Our dyno figures are reference-day numbers at 1.000 bar and 293 K, and the dyno correction calculator explains how to compare them against a real sheet.
Run the turbo six · Browse every turbocharged engine in the roster, filtered by induction
Frequently Asked Questions
Does a bigger turbo always mean more lag?
In our simulator, yes at every speed we tested, and the ordering never flips. At 3,500 rpm on the same 3.0 L six, time to +0.5 bar of boost is 0.10 s with a 60 mm compressor wheel, 0.17 with a 70 mm, 0.33 with the stock 82 mm, 0.82 with a 95 mm, and the 110 mm wheel never gets there. Rotor inertia grows with the fifth power of wheel diameter, ×20.7 across that range, while the turbine's exhaust window grows with the square, and the flywheel wins the transient.
What causes turbo lag?
Rotor inertia, spun up by exhaust energy that is scarce at the moment you ask for it. The compressor shares a shaft with the turbine, so boost cannot rise until exhaust flow has accelerated the whole rotor, and from a closed throttle at low rpm there is little exhaust to do it with (Wikipedia, Turbocharger, retrieved 2026-09-11). On our stock wheel at 2,500 rpm the rotor climbs from 2,016 to 73,180 rpm over a step, and manifold pressure follows the rotor.
Do twin turbos have less lag than a single?
In this model, no: two 59.5 mm wheels reach 90% of their boost in 1.15 s at 2,500 rpm against 0.86 s for one 75.0 mm wheel, at the same power to about 1%. Each small wheel carries 0.315× the inertia, receives half the exhaust through half the throat, and has to spin 1.26× as fast for the same pressure ratio. On a real car twins often win, and the reason is manifold packaging: shorter runners, less volume per turbine and pulse separation, none of which the simulator's single lumped manifold represents.
How do you reduce turbo lag?
Cut the rotor's inertia or add torque to it from somewhere other than the exhaust. Manufacturers use smaller and lighter wheels, variable-geometry turbines, twin or sequential turbos and, on electrically assisted turbos, a motor on the shaft (Wikipedia, Turbocharger, retrieved 2026-09-11). In our simulator the one lever you have is the wheel, and it is a steep one: from the stock 82 mm to 70 mm takes the half-bar wait at 3,500 rpm from 0.33 s to 0.17 s and makes the 1.92 bar target holdable from 2,000 rpm instead of 3,000. The cost is exhaust backpressure, 3.85 bar at redline against 2.73, a rotor at 96.7% of its limit against 83.4%, and a wastegate out of travel from 5,000 rpm up.
Is a small turbo better for daily driving?
Below about 3,000 rpm, on our numbers, a smaller wheel is the one doing the work. At 2,500 rpm the 70 mm wheel holds 2.031 bar, past the target, and makes 482.0 N·m; the stock 82 mm makes 371.8 with the manifold at 1.566 bar, and the 95 mm makes 270.3 with the manifold at 1.152. The trade is at the top: the 70 mm wheel's wastegate is 100% open from 5,000 rpm and the exhaust is at 3.85 bar at redline. The simulator charges that backpressure but not the heat of a small wheel run hard, so read our small-wheel numbers as a direction. The destination is off the model's map.
How These Numbers Were Made
It matters which kind of claim each figure is:
- Physical principles that hold on real hardware: turbo lag as rotor inertia spun by exhaust energy, the moment of inertia of a disc as mass times radius squared and mass as the cube of a length for fixed proportions, the resulting fifth-power law, and the list of ways manufacturers attack lag. Corroborated by the cited sources, not derived from our code.
- Simulator calibration: every horsepower, torque, boost, backpressure, rotor-speed and lag figure above. Rotor inertia is 2.2e-5·(D/0.076)^5 kg·m², the turbine wheel is 0.88× the compressor, the turbine throat and the wastegate port are each a fixed 0.1714 of the turbine wheel's disc (the port then scaled by 0.8), the tip-speed limit is 550 m/s and the target is 1.92 bar. The twin layout sizes each wheel from the displacement it serves and conserves total throat area. The "never" cells belong to this model's eight-second window.
- Method: each configuration built the way the preset builder builds it, with the compressor wheel as the one changed input. Peaks come from 29-point wide-open-throttle sweeps in reference air of 1.000 bar and 293 K with no correction applied. Held points ran wide open at a fixed rpm for six seconds with torque and pressures averaged over the last three and knock read as the maximum. Throttle steps held rpm with the throttle shut for three seconds, then wide open for eight, with final boost taken as the mean over the last second and each threshold recorded at its first crossing. Fuel unlimited, rev limiter ignored, 98 octane as the preset ships.
- The control: the same tests on the simulator build this article was first published from, on 2026-09-11, reproduce every figure published then within one 4 ms sample, so the method did not move and the figures above are the current model's, measured 2026-10-07. The second threshold was +1.0 bar of boost then, when the preset's target was 2.0 bar; at 1.92 bar a full bar of boost is above the target and only a creeping gate reaches it, so it is the target now.
- No real-engine power, torque, boost or lag figure appears above. Where we describe what real hardware does, it carries a citation.
Every run reproduces from the shipped preset with the compressor wheel set from the spec panel, because the knock model's random component is seeded and every figure here repeats. If you land somewhere else, send it to us.
About this article
Written by the Engine Simulator Team, who build and calibrate the physics engine behind these numbers. We publish the model's limits beside its results, because a simulated figure is worth nothing without them. Found an error? Contact the team.
Sources and Further Reading
- Garrett Motion, What is a turbo and how does it work?: the compressor and turbine on one shaft, spun by exhaust energy (retrieved 2026-09-11)
- Wikipedia, Turbocharger: the definition of turbo lag, its cause in rotor inertia and exhaust flow, and the standard methods of reducing it (retrieved 2026-09-11)
- Wikipedia, Compressor map: the surge line at low flow and the choke boundary at high flow that a small wheel runs into (retrieved 2026-09-11)
- Wikipedia, List of moments of inertia: the solid-disc formula behind the fifth-power scaling (retrieved 2026-09-11)
- Wikipedia, Formula One engines: the MGU-H as the motor-generator on the turbocharger shaft between turbine and compressor (retrieved 2026-09-11)
- Wikipedia, 2026 Formula One World Championship: the 2026 power unit regulations and the removal of the MGU-H (retrieved 2026-09-11)
- The Race, Mercedes' new engine upgrade plan is terrible news for Ferrari: Mercedes' 2026 power unit upgrade, with the turbo named as the most obvious area for gains (retrieved 2026-09-11)
Related Reading
- How a turbocharger works: the wastegate, the intercooler, and the single stock-wheel spool trace this article extends
- Supercharger vs turbocharger: which makes more power?: the belt-driven answer to lag, measured the same day
- Roots vs centrifugal superchargers: two compressors with no spool time and a belt bill instead
- How much horsepower one point of compression adds: why this six runs 8.5:1 and what a bare 8.5:1 engine makes below boost
- Compressor stall vs surge: the map walls a real small wheel would hit
- Why F1 engines rev to 20,000 rpm: the engine speeds at which the 2026 power units now spool without electric help
- Browse every engine in the simulator: filter by induction to find the turbocharged presets