Benchmarked against a £25,000 consultancy study

A fire model should not take five months.

A CFD fire strategy study costs around £25,000 and takes five months to return — so it gets commissioned once, late, and the design is arranged around the answer instead of informed by it. Ashbeck returns the same class of analysis in hours, while the design is still yours to change.

Arrange a demonstration → Read a specimen report

Eleven-storey central London tower, rebuilt clause by clause

ashbeck — 25 Marylebone Road / baseline
Project
Drawings
Building
Scope
Simulate
Outputs

Baseline — as designed

1,327,200 cells · 24 meshes · 0.2 m · t_end 1500 s
Running
1080 / 1500 s~18 min remaining
Governing ASET
214 s
Peak pressure
2,729 Pa
Storeys tenable
7 / 8
CONF-01 The deck implements every one of the 8 declarations checked.
MSV-01 Achieved extract 6.9 m³/s against 7.0 declared — within tolerance.
MSV-03 Damper free area 0.67 m² exceeds the 0.60 m² shaft. Reported as declared, not corrected.
PRES-01 Reached 2,729 Pa, above the bound for 4 s, then relieved. Blocks issue until reviewed.
Hours
drawings to a first defensible answer
5 months
what the benchmark study took
£25,000
the fee our pricing answers for
30+
checks that must pass before issue
Everything around the physics

The solver was never the slow part.

It is the fortnight of setup, the queue, the review round, and all of it again for the variation. Ashbeck takes that away.

Describe it in plain language

No deck files, no mesh syntax. You describe the building the way you would to a colleague. Where a value matters and you have not given one, it asks — and where it has to assume, it says so on the face of the report.

It checks its own work

Thirty-plus deterministic checks run against the model and the result. Each names what it examined, what it found and what to do. A blocker stops the document before it exists, not in a footnote afterwards.

Solved in parallel

Divided across meshes and solved simultaneously — work that would otherwise occupy a workstation for days.

Costed before it runs

Runtime and memory stated up front, from a cost model measured on real runs. No surprise at hour forty.

Watched, not babysat

Monitored, hard-deadlined and stopped if they wedge. You are told what happened rather than finding a stalled job on Monday.

In build

Compute on demand

Machines provisioned per job and released after, so a large model never waits for a free workstation. The safety-critical half is already built and tested — tagging, hard deadlines, and the sweep that guarantees nothing is left running and billing.

In build

Progress on your phone

Job status and results wherever you are, with a notification when a run finishes or a check blocks. Push already works; the app is next. We would rather tell you these are coming than imply they are here.

Scoping

It asks the questions a senior engineer would.

Every assumption stays on screen — visible, editable, and quoted in the document. Nothing important is buried in a deck nobody opens.

Engineer

Eleven storeys, two cores, mechanical shaft serving L01 to L08. Office, no sprinklers.

Ashbeck

That is a BS 9999 firefighting shaft study. I have set a 12 MW design fire from BR 368 for an unsprinklered office — say if your fire load differs. I need the shaft cross-section, and whether the lobby doors are held closed at ignition.

Engineer

0.6 m². Doors closed.

Flagged before running Damper free area of 0.67 m² exceeds the shaft cross-section. Passed through as declared rather than silently corrected — it is a finding about a real design.
What to test next

The answer is rarely the end of the question.

When a result comes back short, the useful question is not whether it failed — it is what would change it, and by how much. Ashbeck reads what actually failed in your run and returns the changes worth testing, ordered by their bearing on that failure.

None carries a verdict until it has been run. Whether a scheme is acceptable stays the judgement of the engineer who signs it.

Proposed — not modelled
Raise extract duty by 10%

Stair 1 was not tenable for the whole run. Raises the fire-floor damper rate; check door opening force afterwards.

Proposed — not modelled
Larger damper free area

Lowers face velocity at the damper without revisiting fan selection.

Proposed — not modelled
Robust glazing specification

Delays façade failure, which changes when the compartment vents and therefore when tenability is lost.

Evidence

We rebuilt a £25,000 study and checked our work against it.

Their building, their design fire, their acceptance criteria — compared clause by clause. Fire engineering carries liability, and the failure that costs you is not a tool that stops. It is one that produces a confident, specific, wrong number while every other indicator reads green.

It refuses what it cannot support

If the model sealed itself, or the simulation does not implement what the model declared, you are told before a document exists.

Not measured never reads as fine

Anything the assessment could not evaluate is reported as unevaluated, by name. Silence is never recorded as a pass.

Every figure names its source

Each limit carries the standard it came from, and a citation that cannot be resolved stops the report.

Next step

Bring us a building you want an answer on.

Thirty minutes, an engineer on both sides, and a straight answer about whether this fits the work you do.

Arrange a demonstration → Read the validation dossier