Studio Matrx Monthly · Volume 1 · Issue 3 · August 2026
Amogh N P
 In loving memory of Amogh N P — Architect · Designer · Visionary 
Simulation in the Design ProcessLesson 10.1
BPS for Architecture, Planning & Urban Design/Module 10 · Workflow, Validation & Career

Lesson 10.1 · Workflow, Validation & Career

Simulation in the Design Process

Weaving performance analysis through concept, schematic, detailed and compliance stages

13 min Interactive lessonFree · open lessonByAmogh N P· Architect & interior designer
The hook

A simulation that arrives after the facade is fixed is a report card. One that arrives at concept is a steering wheel.

Most simulation fails not because the physics is wrong but because it arrives too late. The model is beautiful, the numbers are precise, and the design is already frozen - so all the study can do is grade a decision nobody can change.

The fix is not a better solver. It is a workflow: matching the kind of analysis to the stage of design, so that a rough answer lands while the big moves are still open and a precise answer confirms them once they are locked. This lesson is about making simulation part of how you design, not something you do to a design.

Right model, right moment. Deliver before the meeting, not after.

The four stages, and what each needs

A design does not move at one speed, and neither should its analysis. Think of four stages, each asking a different question.

Concept / massing. The questions are the biggest ones: orientation, form, window-to-wall ratio, where the mass goes. Here you want a shoebox model - a handful of simplified zones, default constructions, run in minutes - so you can test ten options before lunch. Tools like Ladybug and Honeybee, or a quick OpenStudio model, are built for this. Precision is almost irrelevant; spread between options is everything.

Schematic. Now the massing is chosen and you are resolving zoning, glazing lines, shading depth and envelope build-ups. The model gains real thermal zones, sensible constructions and schedules. You are comparing refinements, not revolutions.

Detailed design. The model becomes a faithful twin: actual U-values, real glazing (SHGC, VLT), HVAC systems, controls. It supports final sizing conversations with the services engineer and the last envelope trade-offs.

Compliance / documentation. A tightly-specified model built to a rulebook - ECBC, the Eco Niwas Samhita, ASHRAE 90.1 Appendix G, or a rating's reference method - to demonstrate the design meets a target. This model is not for exploring; it is for proving, and its assumptions are dictated by the code, not by you.

The mistake beginners make is treating these as one model that simply grows. In practice the questions change, and so should the tool and the effort. In India this maps cleanly onto real practice: an early Ladybug shoebox to fix orientation and window-to-wall ratio for a composite-climate site; a schematic model to size shading and glazing for the Eco Niwas Samhita's residential envelope targets or ECBC's commercial ones; a detailed model with the services engineer for load sizing; and finally a compliance run to document GRIHA, IGBC or code conformance. Each stage answers what the last could not, and pouring detailed-model effort into a concept-stage question is as wasteful as bringing a shoebox to a compliance submission.

SIMULATION ACROSS THE DESIGN STAGESConceptshoebox modelSchematiczoned modelDetailedfull twin modelCompliancecode modelDesign freedom - decisions still openFAST . ROUGH . MANY OPTIONSSLOW . PRECISE . ONE DESIGN
Zoom
Analysis changes with the design stage. A shoebox model at concept tests the big moves in minutes; the model gains fidelity through schematic and detailed design; a code-dictated model proves compliance at the end. Design freedom - the decisions still open - shrinks left to right, which is why the roughest models carry the most leverage.

Concept = shoebox in minutes. Compliance = code model to the letter. Different jobs.

Fast-early-rough beats slow-late-precise

There is a principle that runs through this whole course and comes to a head here: your influence over performance is highest at the start and falls as the design locks in, while the cost of change rises. Put the two together and the conclusion is unavoidable - the leverage of a simulation is greatest early, exactly when the model is roughest.

This feels backwards to beginners, who want to wait until they have 'enough detail to be accurate.' But accuracy is the wrong target early. A shoebox that says rotating this block 30 degrees cuts cooling load by roughly a fifth is worth more than a month-nine model that computes the annual energy of a frozen design to two decimal places. The first changes the building; the second describes it.

So match the model's precision to the decision at hand. A concept-stage orientation study needs relative honesty across options, not absolute truth. Ask: what is the smallest, fastest model that can separate my options? Reserve the slow, detailed, meticulously-calibrated model for when the decision has narrowed to one design and the remaining question is whether it clears a target. Front-load the cheap studies; they earn the most.

Think of it in leverage terms. Suppose an orientation study at concept, taking an afternoon, reveals that turning the main block to reduce west-facing glazing cuts peak cooling load by roughly a fifth. That single, rough finding may save more energy over the building's life than a month of painstaking envelope optimisation applied later to a poorly-oriented box. The rough study changed the thing that mattered; the precise one polished a thing that did not. This is not an argument against precision - it is an argument for sequencing it correctly. Be rough and broad while the option space is open, and precise and narrow only once it has closed to the design you will actually build.

RIGHT MODEL, RIGHT MOMENTdesign stage: early -> latemodel precisionmatch precision to stageshoebox: rough + earlyhigh value - steers big movesfull model: precise + lateconfirms a frozen design
Zoom
Match the model's precision to the decision on the table. Early in design, a rough shoebox that separates options carries high value; late in design, a precise full model mostly confirms a frozen design. The sweet spot tracks the diagonal - precision should rise only as the decisions narrow.

Influence high + change cheap = simulate NOW, roughly.

Collaborating so results are heard

A model steers a project only if the people making decisions trust it and get it in time. That is a communication problem as much as a technical one.

First, run to the design calendar, not your own. If the orientation is decided at the concept charrette on Tuesday, a brilliant study delivered Thursday is worthless. Time analysis to land before the meeting where the decision is made, even if that means it is rougher.

Second, report differences, not dashboards. Designers and clients do not want twelve KPIs; they want to know which option is better and by how much. 'Option B uses about 18% less cooling energy and holds comfort for more hours - here is why' beats a wall of numbers. Lead with the decision.

Third, make assumptions visible. Say what weather file, occupancy, schedules and set-points you used, and flag the ones that could flip the answer. This protects you and builds trust: when the client asks 'what if we run it 24/7?', you already know.

Finally, stay in the room. The energy modeller who sits with the architect and services engineer, sketching options live, is worth ten who email a PDF a week later. Integrated design - the whole team reasoning about performance together, early - is the setting where simulation does its best work.

There is also a trust dimension worth naming. Designers are rightly wary of a black box that hands down verdicts, so show your working: a short chart of options with the key assumption highlighted invites the team to push back, refine the question and re-run. That back-and-forth is not friction to be minimised - it is the mechanism by which a model actually shapes a building. A result presented as a conversation gets used; a result presented as a ruling gets resented and quietly ignored.

Avoiding simulation-as-afterthought

The anti-pattern is everywhere: the design is essentially complete, someone remembers a rating or a code, and a modeller is hired to 'run the numbers.' At that point simulation can only certify or embarrass - it cannot design. Worse, if the numbers fail, the team is forced into expensive, ugly fixes (more glazing spandrel here, a bigger chiller there) that a concept-stage study would have avoided for free.

Guarding against this is mostly a matter of habit. Book a performance conversation into the concept phase of every project, however short. Keep a living shoebox model that grows with the design rather than a series of throwaway ones. Write down, at each stage, the one or two performance questions that matter next, so analysis has a target. And resist the temptation to over-model: a precise answer to a question nobody is asking is still waste.

Done well, simulation stops being a deliverable and becomes a way of seeing - a habit of asking 'how will this behave?' at every fork in the design, and having a cheap way to find out. That habit, more than any single tool, is what separates a designer who models from a technician who runs software.

One organisational tip makes the habit stick: give the model an owner and a home. A single living file, version-dated, that any team member can open to see the current assumptions and the last set of results, prevents the fragmentation where three people run three private studies that never reconcile. When the model is shared and visible, questions get asked of it continuously - 'what does the model say about moving that wall?' - and it earns its place as a design instrument rather than a report generated once and filed. That is what integration looks like in practice: not a bigger model, but a model the whole team actually reaches for.

Book the performance talk into concept. A living model, not throwaway ones.

Workflow terms & tools in this lesson

Shoebox model

A deliberately simplified early-stage energy model

A few zones, default constructions, runs in minutes - built to compare big moves fast, not to predict absolute energy.

Ladybug Tools

Grasshopper environmental analysis (Ladybug, Honeybee)

Free and iterative; ideal for concept and schematic studies where you re-run as the geometry changes.

Integrated / IDP charrette

The whole design team reasoning about performance together, early

The setting where simulation does its best work; results land before decisions rather than after them.

ASHRAE 90.1 Appendix G

The performance-rating (baseline vs proposed) compliance method

A dictated, code-driven model used to prove a target - exploration is over by this stage.

Hands-on workshop

Workshop - stage a project's analysis plan

Before touching software, plan when each simulation should happen. A good analysis plan is what turns simulation from an afterthought into a design driver.

Paper first. Then free tools as you execute: Ladybug/Honeybee in Rhino/Grasshopper for early studies, OpenStudio/EnergyPlus for detailed and compliance models.

Given & goal
Goal: map performance questions to design stages for a real project
Inputs: a project you know or a studio brief, this lesson's four-stage frame, a notebook
Time: ~30 minutes
  1. 1Write the project's four stages across a page: concept, schematic, detailed, compliance. Under each, note what is actually decided there (e.g. orientation at concept; glazing lines at schematic).
  2. 2For each stage, list the one or two performance questions that must be answered to make those decisions well - phrased comparatively (option A vs B).
  3. 3For each question, name the smallest, fastest model that could answer it, and the free tool you would use (Ladybug shoebox, a quick OpenStudio run, a daylight study).
  4. 4Mark, on a calendar, the design meeting each study must land before. This is the deadline that matters - not when the model is 'finished'.
  5. 5For the compliance stage, note which rulebook applies (ECBC, Eco Niwas Samhita, ASHRAE 90.1, a rating) and that its assumptions are dictated, not chosen.

You’ll walk away with
A one-page analysis plan: four stages, the decisions each holds, the comparative questions, the fast model for each, and the meeting each study feeds. This is a professional's simulation brief.

The worked example

Three altitudes on the same idea

Read the band that fits you — or all three.

For the architectPerformance-driven design decisions

The concept and schematic stages are yours, and they are where performance is won or lost. Orientation, massing, WWR and shading are set before a single detailed model exists - so insist on a fast shoebox study at the charrette, framed as options to compare. Bring the energy modeller into the room early; a five-minute massing test can outweigh months of later refinement.

For the interior designerComfort, daylight & healthy interiors

Your decisions land at schematic and detailed stages, where daylight, glazing and layout are resolved. That is the moment a quick daylight or comfort study earns its keep - arguing for the right glass, the right shade depth, the right desk position in terms of how the space will feel. Time your analysis to the design meeting, and report the difference between options, not a table of numbers.

For the studentSkills, portfolio & green-building jobs

Learn the workflow, not just the software. A studio project that shows a rough concept-stage study steering a design choice - then a refined model confirming it - reads as professional maturity. Practise building a shoebox in minutes and growing it stage by stage; the skill employers value is knowing which model to run when, not which menu hides the run button.

Misconception check

You should wait until the design is detailed before simulating, so the model is accurate.

This gets the value of simulation exactly backwards. Waiting for detail buys accuracy you do not need at the price of leverage you cannot recover. The decisions that dominate performance - site, orientation, form, WWR - are made at concept, when detail is scarce; a rough model that separates those options is worth far more than a precise one that arrives after they are frozen. Accuracy is the goal only at the end, for compliance. Early, the goal is a fast, honest comparison - the spread between options, which is robust even from a crude model. Match the model's fidelity to the decision on the table, and simulate at every stage, not just the last one.
Try it

Do it yourself

Reason it through - no software needed.

  1. 1Name the four design stages and the kind of model each calls for.
  2. 2Why is a rough concept model often worth more than a precise detailed one?
  3. 3What does 'match the model's precision to the decision' mean in practice?
  4. 4Give two ways to make sure a simulation result is actually heard by the design team.
  5. 5What is 'simulation-as-afterthought', and how do you guard against it?
Take this with you

The one line to carry out

Simulation steers a project only when it runs on the design calendar - fast and rough while the big moves are open, precise and narrow once they are locked. Match the model to the decision, deliver before the meeting, and report the difference between options, not a dashboard.
Take it further
References & further reading

Peer-reviewed journals & authoritative standards

  1. 01Hensen, J. L. M. & Lamberts, R. (eds) - Building Performance Simulation for Design and Operation (2nd ed.)Routledge, 2019.
  2. 02IBPSA - International Building Performance Simulation Associationibpsa.org, 2026.
  3. 03EnergyPlus - Whole-building energy simulation engineUS Department of Energy, 2026.
  4. 04OpenStudio - Energy modelling platformNREL, 2026.
Related lessons
Recap
Analysis changes with the stage: shoebox at concept, zoned model at schematic, faithful twin at detailed, code model at compliance. Because influence is highest and change cheapest early, fast-early-rough beats slow-late-precise. And a result only counts if it lands before the decision and is reported as a clear comparison the team trusts.
Carry forward →

Running the right model at the right time gives you a prediction - but how do you know that prediction is any good? Next we close the loop, comparing a model against real metered data and the error bands that say when it can be trusted.

A

The author

Amogh N P

Architect, interior designer, and creative polymath. Studio Matrx began in his notebooks — his vision of design made honest, useful, and open to everyone. Its Academy is written and taught in his memory, and free, forever.

More about Amogh →