The plane decides what can exist, the stack decides how tall

Massing is usually discussed as a single act of judgement, and it produces a shape that everybody then argues about as a whole. Split it and both halves become tractable.

The horizontal question is a packing problem. There is a bounded region. There are objects that have to fit inside it: the hot floor, ward accommodation, outpatients, plant, parking, roads, and the land held for what comes later. Each has a minimum dimension, some have required adjacencies, several have separation distances that must be kept clear, and the objective is to fit them while maximising the room they have to work in. Nothing about floors enters this question at all.

The vertical question is a scalar per object. Given a floor area to accommodate and a footprint the plane can spare, how many floors does this block rise to? That has its own optimum and its own arithmetic, and the answer is not always the lowest number the site permits.

The reason to separate them is that the plane constrains the stack rather than deciding it. A tight parcel forces a building upward, which is why constrained urban hospitals are tall. A generous parcel does not force anything, which means the height becomes a real choice, and a choice nobody has to make is a choice usually made badly.

Start by taking the claims off, because the usable plane is considerably smaller than the parcel and the difference is not small.

The setback removes a band all the way round. Roads, the ambulance approach and the service yards take a substantial area, and on this project they take more than usual, because separating the patient streams means four routes rather than one. Parking is large because this catchment mostly cannot arrive any other way. The plant compound comes with separation distances that are horizontal and not negotiable. And the helipad approach corridors consume plane without occupying it.

The figures here are for a parcel in the middle of the range the land page evaluates, and the interesting thing about them is how little slack there is. Two of the claims have already been compromised to make it fit. Parking has been decked rather than laid out at grade, which is a structure nobody wanted and the cheapest way to buy back plane. And the helipad approach has been trimmed to the point where the pad is arguing itself onto the roof, with the structural premium that implies.

That is the honest shape of the trade. A larger parcel buys surface parking, a ground level helipad, a plant compound that never competes with clinical area, and an expansion reserve that is simply land not yet built on. It buys them with catchment. A smaller parcel keeps the catchment and pays in structures, storeys and compromises, each of which is visible in the capital cost in a way the lost catchment never is.

The last row is the one to watch across that whole range. It is not unused land. It is the horizontal capacity of every decision that has not been taken yet, and it is the only row that can be spent twice, because it can be sold, built on, or left as the thing that lets the hospital still be right in year twenty.

Packing the plane is only half of the horizontal problem. The other half is how each object is divided once it is there, because a building is not a single space and the divisions have their own rules.

Read the last column, because it separates two kinds of limit that are treated identically and behave completely differently. Some of these break loudly. Exceed a fire sub compartment and the evacuation strategy stops working, which is a regulatory failure somebody will catch. Exceed the observation limit in a multi bed room and the infection argument turns, which is at least arguable in a review.

The rest degrade silently. A ward that is slightly too large does not fail. It produces a nurse who walks further and sees less, every shift, and the establishment absorbs it, and no report anywhere will ever attribute the absorption to the plan. Those are the dangerous ones, because there is no moment at which anybody is required to notice.

The fire sub compartment deserves particular attention here and it is the one most often adopted rather than derived. Progressive horizontal evacuation assumes bed bound patients are moved sideways into an adjoining compartment, which holds until help arrives. That strategy is sound everywhere. The compartment size that implements it was calibrated against a response time, and if the response time here is longer then the compartment has to hold longer, which means smaller sections or better construction. It is the clearest case on this site of an imported number that needs re-deriving rather than adopting.

And the least glamorous row is the structural grid. Column spacing that does not divide cleanly into clinical room sizes wastes area at every single column, forever, and on a building of this size that is several per cent of the floor area. It is decided by engineers optimising for something else entirely unless somebody makes it a clinical requirement early.

Now the vertical decision, with the plane settled and the footprint available known.

The instinct on a generous site is to go low, and at the extreme it is available: the whole hospital could be a single storey inside the horizontal capacity computed above. The table says what that costs. A single storey building of this floor area is a hundred and twenty metres square, and the longest journey across it is around two hundred and forty metres, walked repeatedly, by people whose time this study has already priced.

Going up shortens every horizontal journey and introduces a vertical one. The lift adds a wait, a ride and two sets of doors, and both the ride and the share of journeys that need one grow with the building. So one term falls and the other rises, and the sum has a minimum.

It sits at five or six floors, and the interesting part is what that number coincides with. The comparator, on a constrained urban parcel with a planning height limit, is five floors and a basement. It arrived there because it had to. This calculation arrives at the same place with no site constraint at all, which is the useful kind of agreement: the constrained answer and the free answer are the same, so the constraint was not costing the comparator anything on this particular measure.

What it does mean is that a generous parcel does not buy a low building. It buys the absence of a forced tall one, plus room for everything that is not the hospital, plus the expansion reserve. A project that concludes from a roomy site that it should spread out horizontally has misread the table, and one that concludes from a tight site that it must go high has read only half of it.

A basement is a floor like any other, except that the cost of it is paid to the ground rather than to the structure, and the ground on a coastal parcel has an opinion.

The group’s position is that a basement is close to a requirement, and the reasoning is sound enough to state properly rather than treat as a preference. It is the cleanest way to take an entire stream off the plane. Plant, bulk stores, the service route, waste and the mortuary all need space, none of them need daylight, and every one of them is a flow that the flow page wants separated from patients. Separating them horizontally costs frontage, routes and land. Separating them vertically costs one slab. A basement is the only move in the whole massing that removes a set of crossed flows completely rather than scheduling around them.

It also gives shorter vertical service distribution, a mortuary route nobody has to see, thermally stable space in a hot climate, and a plant room that is not occupying the most valuable ground in the site.

What it costs is entirely a function of what is underneath. On coastal Lagos parcels that means a high water table, frequently soft compressible material and in places reclaimed fill, which brings dewatering during construction, permanent tanking against water pressure, a structure designed against uplift, and the single most uncertain line in the capital estimate. It also concentrates the consequences of getting the flood question wrong, because a flooded basement takes with it precisely the plant the hospital cannot operate without.

So the honest reading is not that the basement is unaffordable. It is that the basement is a constraint on the parcel rather than a decision about the building. A site whose ground will take a basement economically is a site on higher, firmer ground, further from the water, which is the same direction the land requirement already points once the area is set above the minimum. Two requirements that arrived separately turn out to select the same kind of parcel, and where that happens it is usually worth following.

And the fallback, if the ground on the chosen parcel turns out to be hostile, is not to abandon the idea but to shrink it. A partial basement under the plant and service zone only, shallow, well drained, with the stores at grade, keeps most of the flow separation for a fraction of the excavation and the risk. That is a decision to take with borehole results in hand rather than in advance of them.

A helipad reads as a rooftop feature and is mostly not. The pad itself is a slab with lighting, drainage, firefighting provision and a structure designed for the load, which is expensive and bounded. The part that constrains the project is the air above the plane.

A helicopter needs approach and departure paths, ideally two from different directions so that the wind cannot close the site, and those paths have to stay clear of obstruction permanently. That converts a piece of the horizontal space into land with a height restriction over it forever, which is why the claim sits on the plane table rather than in the stack.

Which reverses the usual siting logic. On a constrained urban site the roof is the only option, and it carries a structural premium, a dedicated lift, and a fire strategy for an aircraft above an occupied building. On a parcel of this size a ground level pad is available and dramatically cheaper, and it sits closer to the emergency department, which is where the patient is going.

The case for having one at all is not the transfer volumes, which in this market would be low for years. It is trust, and on this site that is not a soft argument. The family testimony concludes that the barrier keeping patients on aeroplanes is evidence and accountability rather than capability, and that the things which would keep them home are mostly cheap and slow, because a promise only becomes worth anything after it has been kept publicly for several years.

A helipad is the exception to the slowness. It is expensive, it is visible, it cannot be faked, and it says one specific thing that a brochure cannot: this place expects the bad night and has built for it. Signals of that kind work precisely because they cost something, and they work on the audience that matters here, which is the referring specialist and the adult child deciding where to send a parent. It is the fastest trust instrument available to this project, and probably the only one that acts in months rather than years.

That comes with a condition, and it is the same condition attached to every signal. It has to be true. A pad that no aircraft ever uses, with no transfer agreements behind it and no protocol for who arrives on it, degrades into decoration and eventually into a joke, and a discredited signal is worse than none because it teaches people to discount the rest. So if it is built it needs the agreements, the training and the published activity to go with it.

The practical conclusion is unchanged and now firmer. Reserve the approach corridors at massing, because they cost almost nothing on the plane and are unavailable once the site has grown. Build the pad early rather than late, because the trust it buys is worth most before the hospital has a reputation of its own. And build it at ground level, because on this parcel there is no reason to pay the rooftop premium.

The two decisions interact, so the order they are taken in matters, and there is a right one.

Settle the plane first. Which objects exist, what shape each must be, where they sit relative to one another, what separation and approach they need, and what is held back. That is the decision with the most constraints in it and the fewest free variables, and it is the one that cannot be revisited, because roads, plant, approach paths and expansion land are all fixed positions on the ground.

Then take the stack per object, within the footprint the plane has allowed. Wards want to be shallow and can be stacked, because a ward floor repeats well. The hot floor wants one large plate and should be low. Outpatients wants to be near the entrance and near the parking, which is a horizontal statement that usually resolves to being on the ground floor rather than a tall block.

Taken the other way round, which is how it usually happens, the height gets set early by a capital cost model or a planning assumption, the footprint follows from it, and the plane is then asked to accommodate whatever is left over. That is how a hospital ends up with its plant in a basement it did not need, its expansion land under a car park, and a helipad approach over the only direction the theatres could have grown.

Three, and the third is the one that applies specifically because the parcel is generous.

A tight site imposes discipline involuntarily. Everything has to be close to everything because there is nowhere else for it to go, and the resulting plan is compact because it could not be anything else. Remove that constraint and nothing stops departments drifting apart, and the walking cost of that drift is permanent and invisible in exactly the way this study keeps describing. The land should buy expansion capacity and resilience. It should not buy slack in the plan, and the only protection against it is to set an internal journey target and hold to it as though the site were still small.