CPA/TCPA and Unpiloted-Vessel Alerts: Collision-Risk Tooling for the Dispatch Board

A dispatch desk has always been a place where the picture arrives late. A pilot calls in a close-quarters situation he is already managing; an agent calls to ask why the inbound has been holding off the sea buoy for forty minutes; the Coast Guard captain-of-the-port calls because a registered vessel is moving in compulsory waters without anyone aboard who is supposed to be there. By the time the VHF crackles, the event is already in progress. The dispatcher is reacting, not anticipating.

That gap is closable. The bridge has had ARPA for fifty years — Closest Point of Approach and Time to Closest Point of Approach computed continuously for every tracked target. There is no reason the shore-side board cannot run the same geometry across the whole district at once, and add a watch the bridge cannot run at all: a check on whether every vessel that is required to have a pilot actually has one. This post is about bringing both of those to the dispatch board.

The two numbers a dispatcher should never have to ask for

CPA and TCPA are the only two numbers that answer the question that matters in a developing meeting situation: will these two pass too close, and if so, when? CPA is the minimum range two vessels will reach if both hold course and speed — measured in nautical miles. TCPA is how many minutes until that closest moment, with a negative value meaning the pair is already opening and there is nothing to act on. We walk through the full vector derivation in CPA and TCPA explained; the short version is that it is a clean dot-product minimization on each pair's relative position and relative velocity.

The reason this belongs on the dispatch board and not just on each bridge is scope. A mate watches the targets his radar has acquired in his own meeting situations. A dispatcher is responsible for an entire district — the bay, the bar, the channel, the anchorages — and cannot watch fifty contacts pair-wise in his head. But the inputs are free: every Class A transponder broadcasts position, SOG, and COG over AIS. A board that ingests the area feed already holds exactly what the CPA/TCPA formula needs, for every moving hull in the zone, with no target acquisition step. The computer can do the n-squared bookkeeping a human cannot.

A concrete alert threshold

Raw collision math without thresholds is noise. In a busy harbor every pair of crossing vessels has some CPA, and surfacing all of them buries the dispatcher under alerts he will learn to ignore within a shift. The board has to be opinionated about what counts as risk. A practical default:

• Red alert when a pair's CPA is under 0.5 nm with a positive TCPA inside 30 minutes.
• Critical escalation at a tighter band — CPA under 0.25 nm — because at that range the situation is no longer developing, it is here.
• A minimum-speed filter so that two vessels swinging on anchor, or a tug holding station, never generate a meeting alert. Two stationary hulls do not have a CPA problem; including them is the fastest way to train people to dismiss the banner.

Those three knobs — distance, time window, and the speed floor — are what separate a useful awareness tool from an alarm that gets muted on day one. The 30-minute window is deliberate: beyond that, courses change, pilots board, traffic services intervene, and a prediction made on present vectors is worth little. Inside 30 minutes, on two hulls both making way, the geometry is meaningful enough to put a name on the board.

And to be clear about what this is and is not: a dispatch-level CPA/TCPA alert is shore-side situational awareness, not a substitute for the vessels' own watchkeeping or for a Vessel Traffic Service. It does not replace COLREGS, it does not direct traffic, and it does not relieve a bridge team of anything. What it does is tell the coordinator assigning pilots and pilot boats that two ships in his district are converging — which is exactly the context he wants before he commits a boat to the water or decides which transit to sequence first.

The watch the bridge cannot run: vessels under way without a pilot

The second piece is one no ARPA unit will ever show you, because it is not a navigation problem — it is a compliance problem, and only the shore side can see it.

Under 46 U.S.C. § 8501, pilotage in the bays, rivers, harbors, and ports of the United States is regulated by state law, and vessels operating under register — foreign-flag ships and U.S. ships in foreign trade — are subject to state pilotage and must, in compulsory waters, embark a state-licensed pilot. A U.S.-flag vessel operating coastwise under enrollment and license falls under the separate federal regime of § 8502 and is generally served by a federally licensed pilot, often a member of its own crew. The dividing line is registry and trade, not size.

Here is where the AIS feed becomes a compliance instrument. Every vessel's transponder broadcasts a static block alongside its dynamic position: type, dimensions, draft, and — critically — flag, derivable from the MID digits of the MMSI. Combine that with the board's own knowledge of which transits it has dispatched a pilot to, and the system can ask a question continuously that no one on a bridge can ask: is there a vessel of compulsory-pilotage character, under way, inside the pilotage boundary, that this association has not put a pilot on?

When the answer is yes, that is a flag worth raising — not as an accusation, but as a prompt. The realistic causes are mundane far more often than they are alarming:

• A vessel that took a pilot from a neighboring association on a shared boundary.
• An exempt operator — certain coastwise units, government vessels, vessels with a valid pilotage exemption certificate.
• A vessel inbound earlier than its declared ETA, getting ahead of the dispatch.
• A genuine gap, where someone needs to make a phone call now.

The board's job is not to adjudicate which of those it is. The board's job is to make sure the dispatcher knows the vessel exists and is moving, before the captain-of-the-port does. A registered tanker stems the channel without a pilot and the first phone call should be the association's, outbound — not the Coast Guard's, inbound.

Why this lives on the dispatch board specifically

You could, in principle, watch all of this on a generic vessel-tracking website with two monitors and infinite attention. In practice nobody can, and the generic tools do not know which transits your association has pilots on, so they cannot run the unpiloted-vessel check at all. The CPA/TCPA banner and the unpiloted-vessel flag are only useful when they sit on the same board that holds the duty rotation, the ETAs to the boarding area, and the transit log — because that is the board where the dispatcher is already looking and already acting.

That integration is the whole point of Binnacle Passage. The collision-risk engine and the compulsory-pilotage watch are not standalone alarm boxes bolted to the side; they read the same live AIS feed that drives the board, cross-referenced against the same dispatch state — who is assigned, who is en route to which boarding point, which transits are open — that the desk maintains anyway. The alert names both vessels, gives the predicted CPA and the minutes to it, and sits next to the rotation so the dispatcher can act in one place.

The same philosophy runs through the rest of the timing tooling. A converging-pair alert tells you that two hulls will meet; deciding whether an inbound can even make the boarding window often comes down to tide and draft, which is why the board pairs the collision picture with timing math — the kind of under-keel and tide-gate reasoning you can try standalone in the squat calculator and the tide window calculator. Awareness and timing are the same job from two angles.

This matters most exactly where the picture is hardest to hold by hand. In a dense, restricted corridor like the Houston Ship Channel, meeting and overtaking situations are continuous and no human can watch them pair-wise across the whole channel at once — the automation earns its keep precisely because there are too many targets to track. The collision engine sorts the genuinely close-quarters situations out of a crowded board so the dispatcher spends attention where the geometry says to.

What it changes on the desk

Put plainly, the change is one of order. Today, the sequence in too many districts is: event happens, VHF call comes in, dispatcher reacts. With the collision and compulsory-pilotage tooling running on the board, the sequence inverts: the board surfaces the converging pair or the unpiloted hull, the dispatcher sees it, and the action — a call, a boat, a sequencing decision — happens before the situation forces it. The radio call, when it comes, confirms what the desk already knew instead of informing it.

That is the entire value proposition. Not more data — dispatchers already drown in contacts — but the right alert, computed automatically, on the board where the decision gets made, early enough to be a decision rather than a reaction.

See both engines on a working board with live traffic: Binnacle Passage runs CPA/TCPA across every moving pair in the zone and flags compulsory-pilotage vessels under way without a pilot, alongside the duty rotation, boarding-point ETAs, and the transit log. Walk through it on the see a live pilot board demo, then tell us which district's water you dispatch — the thresholds are meant to be tuned to it.