Expedition Navigation Software: What We're Looking For

This is a research note, not a rated review. We have not yet run Expedition through a full season on our own campaign, so the assessments below are framed as what the software is designed to do, the engineering underneath it, and what we would be testing.

Expedition is the laptop-based navigation and performance software behind the majority of Grand Prix and offshore programmes, and its single job is to convert a calibrated instrument feed into decisions. Developed since the mid-1990s by physicist and Volvo Ocean Race navigator Nick White, it sits on a Windows machine, consumes the boat's data over NMEA or a network socket, and layers the wind triangle, leeway and tide vectors, polar interpolation, isochrone routing and start-line geometry on top. It is a computation engine, not a chartplotter with extras, and it exposes the truth of your calibration mercilessly.

What Expedition actually computes

Strip away the interface and Expedition is doing four classes of calculation, all downstream of the same raw inputs — speed through the water (STW), apparent wind angle and speed (AWA/AWS), heading, heel and GPS position.

True wind and the wind triangle. Expedition does not receive true wind; it derives it. Apparent wind is the vector sum of the true wind and the boat's own velocity, so true wind speed comes from the law of cosines, TWS = √(STW² + AWS² − 2·STW·AWS·cos AWA), with true wind angle from the companion sine relation. This is where the subtlety lives: the wind is measured relative to heading, but in any tide the boat actually travels along its course over ground (COG), which does not equal heading. The moment you are being set, the naive triangle produces a wrong true wind direction unless the software resolves the current vector (set and drift) and the leeway angle into the solution first. Expedition can be fed STW or work in a ground frame; getting that frame right is the difference between a stable true wind direction and one that wanders every time the tide turns.

Leeway. No sensor measures leeway directly, so it is modelled. The standard relationship, in use since the early 1980s velocity-analysis work, is λ = K · heel / STW², with λ and heel in degrees and STW in knots — leeway grows with heel (more side force) and collapses as the square of speed (the keel makes lift more efficiently). Typical K sits around 10 to 12, and on a powered-up boat leeway is 3 to 7 degrees of real angle that must be subtracted from heading to get the true course through the water before the wind triangle is solved. Better systems replace the single K with a matrix across the wind range; a single constant is a reasonable first approximation and no more.

Targets from the polar. Expedition reads the boat's polar — a table of predicted STW for every true wind speed and angle, itself the output of a velocity prediction program (VPP) that balances driving force, side force and righting moment. Upwind and downwind targets come from the tangent construction: the optimum beating angle and target speed are the point where a line drawn straight up the course direction is tangent to the polar curve (maximising the up-course component, VMG = STW · cos TWA); the run angle is the same construction on the bottom lobe. Between the tabulated points Expedition interpolates, and it reports live target boat speed and target VMG angle so trim and helm can see, in real time, the gap to potential.

Weather and isochrone routing. For offshore and distance work Expedition ingests GRIB1 and GRIB2 fields from Saildocs, PredictWind, the NOAA models and others, merges datasets and runs isochrone optimisation. The isochrone method is a recursive branching search: from the start it projects every feasible heading forward one time step, using the polar to convert forecast wind into achievable speed, draws the isochrone (the front of furthest-reached points at that time), prunes back to the outer envelope, and repeats. The isochrone interval defaults to three hours; shortening it improves resolution around wind shifts and coastlines at the cost of compute. Expedition ships two independent routing algorithms that should converge on near-identical routes — running both and comparing is a cheap sanity check — and applies wave and current corrections and limits on wind speed and wave height.

Underneath all four is data logging at roughly one-second resolution. That log is not a convenience; it is the raw material for validating the polar and the calibration after racing, which is where the real gains compound.

Why the calibrated feed matters more than the software

Expedition measures nothing. It consumes what the instrument system outputs, so its accuracy is capped by the accuracy of four channels — STW, AWA, AWS and heading — plus the polar. Every error propagates: a heading offset skews true wind direction and therefore every layline and shift call; a boat-speed error corrupts both the true-wind solve and the target comparison. The software presents the corrupted answer with total confidence, which is exactly how it erodes the afterguard's trust in the whole system. The work, then, is upstream, in the correction tables the processor applies before Expedition ever sees the data.

Boat speed (STW). Paddlewheel and even electromagnetic sensors are non-linear and site-dependent. The discipline is a measured-mile or two-GPS-waypoint run on a constant compass heading (not COG, so tide cancels over reciprocal runs), at steady revs and speed, three runs each way to average. Paddlewheels in particular tend to progressively over-read at higher speed as the hull boundary layer changes, so a single scale factor is not enough — the correction is a curve across the speed range, and it should be re-checked when the transducer or its fairing is disturbed.

True wind angle — upwash. The masthead vane sits in air already bent by the sails' circulation. On the wind, upwash rotates the local flow toward the sail, so the measured AWA — and the derived TWA — reads larger than reality, and by different amounts on each tack because of wind shear with height. The diagnostic is simple: sail upwind on both tacks and watch computed true wind direction. If it is not identical tack-to-tack you have residual error. The correction is a TWA-versus-TWS table, applied so that if the true wind direction consistently lifts as you tack you subtract half the split, and if it consistently heads you add half — iterating until the direction holds steady across a tack. This table is conditions-dependent and has to be revisited on the water, not set once and forgotten.

True wind speed — downwind over-read. Sailing deep, air accelerates over the rig and up to the masthead unit, so TWS over-reads downwind. The practical correction is on the order of +12 to −15 percent applied to true wind speed on the run; without it, downwind targets and routing weightings are built on an inflated breeze.

The polar. On a strict one-design a class or shared-campaign polar is a legitimate starting point, but it is generic. Winning teams treat the polar as a living file: they overlay logged STW against target, mode by mode and pressure by pressure, trim the numbers where the boat consistently beats or misses the table, and re-baseline over a season. A stale or generic polar is as damaging as bad calibration — it hands the crew targets the boat can never reach, or beats too easily, and either way the number stops meaning anything. "Rubbish in, rubbish out" is not a caution here; it is the governing equation.

How it connects to the boat

Expedition reads the instrument network three ways: NMEA 0183 serial, NMEA 2000 through a gateway, or an Ethernet feed carrying NMEA sentences over TCP or UDP, conventionally on port 10110 (the de facto standard for NMEA-0183-over-IP). The trade-off between the transports is real: TCP is a single acknowledged client-server link — sentences are checked and re-sent if lost, which is what you want for the primary navigator's feed — while UDP broadcasts to every device on the network with no acknowledgement or re-send, which suits pushing data out to several tablets but drops packets silently on a busy or marginal link. Expedition is configured under the Instruments panel as a network or serial connection, with the specific sentences enabled explicitly.

It is worth stressing what Expedition is not: it does not replace a B&G, NKE or WTP processor, and it does not perform the sensor calibration those processors carry. It is the computation and tactical layer that sits on top of whatever instrument system already runs. That single fact drives the most common architecture decision in this space — which software your processor forces on you.

Expedition, Deckman or Adrena

For a serious campaign there are effectively three tools, and the choice is often made by the processor rather than by preference. Deckman remains the most powerful and customisable professional option — teams can write their own add-ins — and a B&G WTP system effectively requires it to communicate at full fidelity; its interface is dated and its curve steep. Adrena is the favourite of many offshore and IMOCA programmes, strong on cartography, weather and strategy, though lighter on some inshore start-line and what-if features. Expedition is the broad default for Grand Prix and mixed inshore-offshore work because it combines the strongest GRIB and multi-model support, two capable routing engines, wide instrument compatibility and the most usable interface of the three, and it is generally the most affordable with updates included. None of the three is casual; all reward a serious time commitment. See our Grand Prix instruments guide and the broader race-boat electronics overview for how the processor choice cascades into everything downstream.

What it looks like on a Melges 40 campaign

The Melges 40 is an all-carbon, strict one-design, canting-keel Grand Prix boat raced windward-leeward against a level fleet, and on that platform the value of Expedition shifts away from offshore weather routing and toward precision under load. Public class material describes a roughly 12 metre boat of about 3,200 kilograms with a push-button canting keel (canting to around 45 degrees) working with a centreline canard, twin rudders, a retractable bowsprit and a tall, powerful rig, quoted at 20-plus knots downwind — figures we would verify against the class rules and the boat's own documentation before relying on any of them, and never treat as a substitute for the measured polar. Two consequences flow directly to the software. First, the canting keel and canard change the righting-moment and leeway behaviour versus a fixed-keel boat, so the leeway K (or K-matrix) and the polar must reflect this configuration, not a generic 40-footer. Second, the acceleration is violent, so live targets are read in a fast-moving pressure band where the gap between target and actual opens and closes in seconds.

Around the cans, races are decided in three places and Expedition speaks to all three. At the start, line bias, distance-to-line and time-to-burn give the afterguard an objective read on the favoured end and whether the boat is on schedule for a full-speed, on-time entry without risking OCS — in a fleet of identical boats, that margin is frequently the race. On the beats and runs, live target boat speed and target VMG tell the trimmers whether the boat is at its polar potential for the pressure of the moment, and the laylines and time-to-mark kill the classic overstand-or-tack-short errors — though on a boat making low-20s downwind, layline geometry is unforgiving and only as good as the calibrated TWA feeding it. And after racing, the one-second log is the debrief: actual versus polar, mode by mode, feeding the season-long refinement of the campaign's own polar so every subsequent target is truer.

Offshore or on distance races the isochrone routing becomes primary; for a Med-circuit one-design programme the honest emphasis is start-line geometry, laylines, live targets and the logged data.

Failure modes, and what good looks like

The ways Expedition disappoints are predictable and all sit in the inputs. Uncalibrated data is the biggest — confident but wrong numbers that quietly destroy trust in the system, most visibly as a true wind direction that splits tack-to-tack. A stale or generic polar yields targets the boat cannot hit or beats too easily, equally useless. No clear owner leaves the calibration tables and polar half-maintained and the laptop a distracting screen nobody fully believes. And fragile plumbing — a marginal UDP link dropping packets, or a laptop that sleeps mid-race — removes the tool exactly when it is needed.

Good looks like the inverse of all four. One person owns the software, the polar and the correction tables end to end: compass swung, boat speed linearised on a measured mile, TWA table built until true wind direction holds across a tack, TWS downwind correction dialled, leeway K set for this boat. The data path is a tested, acknowledged TCP link with the laptop's sleep and USB power management disabled. The polar is validated against the log and refined over the season. And the numbers are trusted enough that the afterguard acts on them by reflex — a time-to-burn call at the gun, a target-speed nudge on the beat, a layline that holds. Formal instruction shortens that road materially: North U and others run dedicated Expedition courses, and the curve is far steeper by trial and error.

Our assessment, honestly framed

We would assess Expedition on exactly the terms above: how cleanly it integrates with the instrument processor and transport; how trustworthy the calibrated STW, wind and heading channels are once the correction tables are built; how stable the derived true wind direction stays tack-to-tack; how usable the start-line and layline mathematics are under Melges 40 loads; and how much genuine insight the one-second log yields in the debrief. We would also weigh who owns the workflow, because on a Grand Prix campaign that single ownership is the line between an advantage and an expensive habit. When we have run it in earnest across a season — with our own validated polar and calibration behind it — we will share rated, specific findings rather than the framing here. For terminology, our sailing terms glossary covers the polar, VMG and layline concepts referenced throughout.