Read this before you copy anything out of this page.
Every number quoted here from a standard, a test report or a published specification is included to show you the shape of a clause, not to give you a value to paste. Threshold values belong to the edition your project adopts, to the specimen that was tested, and to the authority that has jurisdiction where you are building. Do not lift a class, a pressure, a grade or a band from this article into a project manual. Ask for the report behind it.
Search for casement window specification text and you will find plenty of it: a tidy paragraph about thermally broken aluminium frames, multi-point locking, double glazing and powder-coated finishes. It reads well. It is also, almost without exception, unenforceable — not because it is wrong, but because there is no sentence in it that anyone can fail.
That distinction matters more than any parameter in this article. A description tells you what a window is like. A specification is a set of sentences that a named person will later have to prove or disprove, using a named document, at a named moment. If a sentence cannot be failed, it is not a clause. It is a preference that wandered into the contract documents wearing a clause's clothes.
We make the window, so we read these documents from the receiving end. We know exactly which sentences we can answer from a catalogue and which ones send us back to a test house, a calculation, or a conversation. That asymmetry is the subject here. Below is the clause-by-clause anatomy of an aluminium casement window specification: what each part is for, which words quietly transfer risk, why a performance class is a sentence rather than a number, and the clauses that a casement window spec routinely leaves out.
Take any line you are about to write and try to complete four columns for it. If you cannot fill the last three, the first one is decoration.
| Column | The question it answers | What an unfilled column looks like |
|---|---|---|
| The requirement itself | What must be true | "High quality hardware" — true of everything anybody is willing to ship |
| The verifier | Who has to demonstrate it | "The contractor" — the contractor does not make the window |
| The document | What piece of paper settles it | "Test reports" — of which specimen, tested to which edition |
| The moment | When it must exist | No date at all — so it can be answered after handover |
Now write the same requirement as a clause and watch the three empty columns fill themselves. "Hardware: manufacturer's published cycle test report for the exact operator model supplied, submitted with the shop drawings, ten working days before fabrication release." Same intent. One of them can be failed on a Tuesday morning.
There is a second reason the four-column test is worth the trouble. On a project where the specification is descriptive — where it prescribes materials and methods rather than results — the design liability sits with whoever wrote the description. Under the doctrine that grew out of United States v. Spearin in United States law, an owner impliedly warrants that a contractor who builds exactly to a prescriptive specification will get a workable result. That is the trade: a descriptive clause gives you control and hands you the consequence. A performance clause gives the supplier freedom and hands them the burden. Most casement window sections mix the two without noticing which lines are which.
In a MasterFormat-style project manual, aluminium windows sit in Division 08 (Openings), section 08 51 13 Aluminum Windows, written in the three-part SectionFormat: Part 1 General (scope, references, submittals, quality assurance, delivery and storage, warranty), Part 2 Products (materials, fabrication, finishes, hardware, accessories), and Part 3 Execution (examination, preparation, installation, field quality control, adjustment, cleaning, protection). Division 01 General Requirements sits above every technical section and governs the administrative procedures that all of them must follow.
The interesting part is not the three parts. It is the related-sections list, because that list is where casement window failures actually live.
| Related section | What it owns | Why the window section cannot fix it |
|---|---|---|
| 08 80 00 Glazing | The glass make-up, coating, spacer, gas fill | "Double glazed" in 08 51 13 with a different spec in 08 80 00 is two answers to one question |
| 07 92 00 Joint Sealants | The perimeter seal between frame and wall | Most "the window leaks" claims are this line, and it is in Division 07 |
| 07 60 00 Flashing and Sheet Metal | Head flashing, sill pan, the path water takes out | A perfect window in an unflashed opening is a wet window |
| 07 27 00 Air Barriers | Continuity of the air line around the opening | Air leakage measured on site is usually the joint, not the sash |
| 09 06 00 Schedule for Finishes | Colour and coating designation | Naming a colour in two places invites two batches |
| 05 50 00 Metal Fabrications | Steel subframes and reinforcing | A subframe can bridge the thermal break the window was bought for |
The practical consequence: a casement window specification can be fully compliant and the installed window can still leak, because the two documents that decide whether it leaks are not the document titled "Aluminum Windows". When you write or review the section, read the four related sections at the same sitting. And note the drafting convention that governs all of them — drawings show quantity, location and configuration; specifications state quality, material and workmanship. Quality requirements written onto a drawing become a conflict the moment the specification is revised.
One more structural note. A clause set written for one product family does not travel. The clauses you would hand to an aluminum alloy folding door wholesaler — panel stacking, top-hung load paths, folding stack clearance — are not the clauses you hand to a casement supplier, where the load path runs through hinges and a frame that has to resist racking. If you maintain a master specification, keep one section per operating type and resist the urge to merge them.
There are four recognised methods of specifying, and a good casement window section uses all four in different lines. The mistake is not picking the wrong one — it is not noticing which one you picked.
| Method | What it does | Where the risk lands | How it fails in a casement section |
|---|---|---|---|
| Descriptive | States properties, dimensions and workmanship; names no brand | Stays with the specifier | You have now designed the window. If it flexes, the description was yours |
| Reference standard | Cites a published standard by designation | Shared, but you own the citation | "Complies with GB/T 8478" imports every table in it, including the ones you never read |
| Performance | States the required result and how compliance is verified | Moves to the supplier or fabricator | Only works if you also define the specimen — otherwise every bidder tests a different window |
| Proprietary | Names manufacturer, brand, model | Shared once you add "or approved equal" | "Or approved equal" with no criteria is the sentence that buys the worst acceptable window |
Two of these deserve a closer look because they cause the most damage in window sections specifically.
The standard drafting position on reference standards is blunt: provisions incorporated by reference "become a part of the specifications just as though included in their entirety." So when a section says the windows shall comply with a national aluminium window standard, it has just adopted that standard's scope, its classification tables, its tolerances, its test sequences and its sampling rules — including the parts nobody on the design team opened.
That is fine, provided you do two things. First, name the edition. Standards are revised, and the revised version usually moves the numbers. Publicly circulating summaries of one widely used Chinese test method give the same edition's top air-permeability band at two different values, which is a useful reminder: if the popular tables disagree, the only defensible source is the standard text plus the report. Second, name the class, not the standard. A standard contains levels. Citing the document without the level is not a requirement; it is a reading list.
The open proprietary method — naming a product and allowing substitutions — is standard practice and, on many public projects, a competitive-bidding requirement. It works when the substitution clause lists what a substitute must demonstrate. It fails when it does not, because "equal" then means whatever the person under schedule pressure decides it means that week.
The fix is a single sentence appended to the named product: substitutions considered only where the requesting party submits, before bid, documented evidence of compliance with each performance requirement listed in this section. That turns a comparison of brochures into a comparison of documents. A related and often better option is the basis-of-design approach: the named product sets the benchmark, and any alternative must measure up to it in measurable terms. That only works if the benchmark product's data is itself measurable — which brings us to the largest gap in most casement window sections.
Specification language has a small vocabulary with large consequences, and the conventions are taught consistently enough to be worth stating plainly. In the CSI Construction Specifications Practice Guide, "shall" and "will" are used as imperatives for work required of the contractor, while "will" is reserved for acts required of the owner or the architect/engineer; "must" and "is to" are not recommended. Black's Law Dictionary gives "shall" its mandatory sense: has a duty to, is required to — the sense drafters intend and courts typically uphold.
| Word | Usual meaning | What happens when it appears in a window clause |
|---|---|---|
| Shall | Mandatory obligation on the contractor | The line is enforceable. Make sure the supplier is actually a party to it |
| Will | An act by the owner or the A/E | "The manufacturer will provide drawings" obliges nobody who signed the contract |
| Should | Advisory; no obligation attached | The single most expensive word in a specification. delete it or upgrade it |
| May | Permissible, at the option of the party | "Screens may be factory glazed" is not a requirement, it is a shrug |
Alongside these sit a family of adjectives that specification guides tell writers to avoid outright, and they are the ones that show up most often in window sections: good workmanship, adequate, reasonable, satisfactory, generally, workmanlike, best commercial quality, standard practice. None can be measured, so none can be failed, and each one silently converts a requirement into an argument. Two more drafting habits belong in the same bin: writing "as approved", "as required" or "as directed" without naming who approves; and using "and/or", which has no definite meaning.
The positive version is short. Write in the imperative mood — "Provide", "Submit", "Test" — rather than "the contractor shall". Use streamlining: put the subject first and follow it with the requirement, as in Frame depth: 116 mm. Operator cycles: submit test report. Short lines get read; long ones get skimmed, and the skim is where disputes start.
Here is the part of a casement window specification that is most often written wrong, and it is wrong in the same way in every market.
A performance designation looks like a short code. It is not. Read one from an actual published test report and you can see how much is packed into it: the primary designator reads Class AW – PG100 : Size tested 900 x 1500 mm – Type C, and it is accompanied by a positive design pressure, a negative design pressure, a water penetration resistance test pressure, an air infiltration level, a test completion date and a report number. The tested size is inside the designation. That is not a footnote; it is half the meaning. A second published spec section for an inswing casement carries the designation AW-PG80-C, where the trailing letter identifies the operating type.
Two consequences follow, and both belong in your section text.
First, the grade is attached to a specimen, so the specimen must be described. The guidance from the Fenestration and Glazing Industry Alliance is explicit that a manufacturer intending to certify a product in larger sizes must test the largest size it intends to certify and label. It also distinguishes a performance grade from a design pressure, precisely because products were being promoted on structural pressure alone while the other required categories went unmentioned. Clause-level translation: do not accept a pressure figure without the grade, and do not accept the grade without the size and the type.
Second, the edition lags the publication. A newer edition of the North American fenestration standard was published in 2026, while the model codes in force referenced an earlier edition. If your section cites the standard by name without a year, and your authority cites a different year, you have two specifications and one window.
Under EN 14351-1, the harmonised product standard for windows and external doorsets, the manufacturer issues a Declaration of Performance. Read one and you will see that it is a table, not a number. Each row is an opening function — the way the leaf actually opens — and each column carries a declared value, the classification standard behind it, the test standard, the notified body, the test report reference, and the tested size in millimetres.
A real declaration for a triple-glazed door product line shows the single-leaf outward-opening version declared at wind load Class C4 (1600 Pa), watertightness Class 9A (600 Pa) and air permeability Class 4, tested at 950 x 2185 mm. The two-leaf version of the same product line sits directly below it and reads NPD on all three. NPD means No Performance Determined. It is a legal status, not a failure to test — but on a tall facade it is also the honest answer to "what is the rating of the double version", and it is the answer you would never have guessed from the single-leaf figure.
So the clause you write is not "shall be Class 4". It is: air permeability shall be Class 4 to EN 12207, demonstrated by a test report to EN 1026 for the exact opening function, configuration and tested size supplied. And you need to know what the scale means to know whether Class 4 is even the right ask.
| Characteristic | Scale | What the scale actually steps through |
|---|---|---|
| Air permeability, EN 12207 | Class 1 to Class 4 | Higher is tighter; Class 4 is verified up to 600 Pa under EN 1026 |
| Watertightness, EN 12208 | 1A to 9A, then Exxx | Pressure in steps to 600 Pa at 9A; "A" means tested unprotected, the severe method |
| Wind resistance, EN 12210 | Class 1 to 5, then Exxxx, with suffix A/B/C | Number is the load (Class 4 is 1600 Pa); letter is the permitted deflection, A loosest to C stiffest |
| Thermal transmittance | A continuous value, not a class | Computed under EN ISO 10077 or measured under NFRC; boundary conditions differ, so conversions are arithmetic, not equivalences |
The Chinese framework works the same way with different labels. GB/T 7106 sets out the test methods for air permeability, watertightness and wind load resistance of external windows and doors, and the classification runs in grades rather than classes — air permeability in eight grades, watertightness in six, wind load resistance in nine, the last keyed to a P3 value. The lesson for the spec writer is identical: the grade number is meaningless without the report, and the report is meaningless without the specimen description.
One pattern is worth carrying across all three systems. Casement and tilt-turn sashes, which compress a gasket all the way around, typically land higher on air permeability than sliding sashes, which can only wipe. If your section is written around a casement window and then value-engineered into a slider, the air number in the spec is no longer the air number of the product. Write the operating type into the performance clause, not just into the schedule.
Most published casement window specification text covers profile, glass, hardware and finish. Those are the easy four. The following are the ones that decide outcomes, and they are frequently absent.
"Left hand" is two different windows depending on which side of the wall you are standing on. Published casement sections that get this right define it explicitly: hinges on the left when standing outside and looking in. Without the viewpoint, the clause is a coin toss resolved at the Site, by whoever unpacks the crate. The same ambiguity is why any specification covering a left hand swing door has to fix the viewing direction before it fixes the word. Put the viewpoint in the clause. Put handing on the schedule too, per opening, not once in Part 2.
"Casement" tells you the sash swings. It does not tell you whether the unit is vent-to-vent, vent-to-fixed, vent-under-fixed or vent-over-fixed, nor whether the mullion is integral to one master frame. Those are separate clauses with separate consequences for sightline, for structural depth and for how many lock points the unit actually gets. In our own catalogue the mullion is published as a pair of numbers — for example 75/31 on the C120, 75/21 on the C113 and 58/22 on the C116 — where the first number is depth and the second is the visible face. A clause that quotes only the second number is specifying the view and not the structure.
Published aluminium window sections contain two sentences that do more work than any thickness figure, and they are badly under-used:
Both are enforceable on site with a screwdriver and a section drawing, which is exactly why they are worth having. A thermally broken frame fixed with a continuous steel angle across the break is a thermally broken frame that is not thermally broken.
One published institutional specification defines the term that decides most window disputes: uncontrolled water is water not drained to the exterior, or water appearing on the room side of the window. Adopt something equivalent. Without it, "it leaks" is a feeling; with it, "it leaks" is a condition with a pass and a fail. Note what the definition gracefully handles: water sitting in a frame cavity that drains to the outside is doing what it was designed to do. Water on the room side is not.
Cycle figures are test-rig results, not service lives, and a handle cycled on a bench is not a loaded sash cycled in a frame. Both statements are true and both belong in the clause: ask for the report, the standard it was run to, and the configuration of the test piece. Also specify who performs the first adjustment. A multi-point casement that has never been adjusted is not a defective casement; it is an unfinished one, and the sentence allocating that task prevents a commissioning argument eighteen months later.
Where a casement window carries an integrated insect screen, the screen is a second operable leaf with its own cycle count, and it is operated every time the glass is operated. It is also frequently omitted from the hardware clause entirely. Opening limiters and fall-arrest devices belong in the same place: they are safety components, they are cheapbefore fabrication and awkward after it, and in some jurisdictions their presence is a regulatory matter rather than a preference. Check yours with the authority having jurisdiction — we will not endorse a rule for a market we do not build in.
Published institutional sections require a manufacturer's certification that the windows delivered are identical to the windows tested, and they commonly put a freshness limit on the test reports behind that certification. Whether or not you adopt those exact limits, adopt the mechanism: the delivered product must be traceable to the tested product, in writing, before release for fabrication. That one sentence converts a test report from marketing collateral into a contract document.
Site water testing is usually run to the AAMA 502 method, and published sections show the common practice of testing at a pressure above two-thirds of the laboratory pressure. Two things follow. The fraction is a default, not a law — a specifier can require the full laboratory value in writing, and on exposed elevations it is worth doing. And the pre-check matters as much as the test: published sections instruct that a unit outside a tolerance of roughly 3 mm out of plumb, level or square should not be tested at all, because you would then be testing the installation rather than the window. Write both sentences in.
Below is the shape of a casement window section. Every value is a blank, deliberately: the numbers belong to your project, your climate, your code edition and your tested specimen. What is not blank is the structure, the verbs and the four-column discipline.
PART 1 — GENERAL
1.1 Summary. Section includes aluminium casement windows, outward opening, complete with frames, sashes, hardware, glazing, screens and anchorage. Related sections: 07 27 00, 07 60 00, 07 92 00, 08 80 00, 09 06 00.
1.2 References. List each standard cited in this section, with designation, title and edition year. Do not list a standard that is not cited in the body.
1.3 Performance requirements. Air permeability: Class ____ to ____, demonstrated at a tested size of ____ x ____ mm, opening function ____. Watertightness: Class ____ to ____, at ____ Pa. Wind load resistance: Class ____ to ____, deflection suffix ____. Each value requires a report number, a test standard and a tested size.
1.4 Submittals. Shop drawings with section cuts at ____ scale showing the drainage path, the thermal break and the corner construction; hardware schedule by opening; test reports per 1.3; finish samples; certificate of identity to tested specimen; installation and maintenance instructions.
1.5 Quality assurance. Installer qualifications: ____. Mock-up: ____ opening(s), retained as the workmanship standard. Pre-installation meeting: mandatory, attended by the installer, the glazing trade and the sealant trade.
1.6 Delivery, storage and handling. Store upright, under cover, off the ground. Do not stack flat. Protect finishes from contact with wet cementitious materials.
PART 2 — PRODUCTS
2.1 Basis of design. Manufacturer: ____. Series: ____. Substitutions: considered only where the requesting party submits, before bid, documented evidence of compliance with each requirement in 1.3 and 2.2 through 2.7.
2.2 Frames and sashes. Aluminium alloy and temper: ____. Minimum wall thickness of load-bearing members: ____ mm, measured on the metal excluding any coating, and stated for named members. Thermal break: material ____, width ____ mm, multi-chamber.
2.3 Thermal break integrity. The thermal break shall not be compromised by hardware or metal fasteners. Subsills and anchor clips shall not span the thermal break or connect the separated components.
2.4 Glazing. Make-up: ____ mm / cavity ____ mm / ____ mm. Spacer: ____. Coating: ____. Gas fill: ____, with initial content certified in writing. Glass as specified in Section 08 80 00; where this section and 08 80 00 differ, submit the conflict in writing before fabrication.
2.5 Hardware. Operator: ____. Lock points: ____ per sash. Hinge material and rated capacity per sash: ____. Cycle test: submit report to ____ for the model supplied. First adjustment performed by ____, recorded, before handover.
2.6 Configuration and handing. Per schedule. Handing defined as viewed from ____ looking ____. Configuration: ____.
2.7 Screens and safety devices. Screen mesh: ____. Screen operation: ____. Opening limiter: ____. Fall-arrest device: ____ where required by the authority having jurisdiction.
PART 3 — EXECUTION
3.1 Examination. Verify openings are dimensionally correct and within allowable tolerance, and that substrates are plumb, level and clean. Do not proceed until unacceptable conditions are remedied in writing.
3.2 Installation. Install plumb, level and square. Anchor per approved shop drawings. Do not bridge the thermal break. Coordinate perimeter sealing with Section 07 92 00 and flashing with Section 07 60 00. Leave drainage paths unobstructed.
3.3 Field quality control. Field water test to ____ on ____ percent of openings, at ____ percent of the laboratory test pressure. Do not test units outside ____ mm out of plumb, level or square. Retest after correction at no cost to the owner.
3.4 Adjustment and closeout. Adjust all operating sashes. Demonstrate operation, locking and drainage. Submit maintenance instructions and the record of adjustments.
Notice what the skeleton does with dimensions. It asks for frame depth, member thickness and thermal break width as separate blanks, because they are separate properties. A catalogue that offers "113" and "116" is offering frame depth figures; the number in a product name is a nominal call-out, in exactly the way a 36 x 80 entry door is a nominal call-out rather than a manufacturable dimension. Your order needs the manufacturable one.
| The clause | Why nothing can fail it | The rewrite |
|---|---|---|
| Thermally broken aluminium | Says nothing about the material or the width of the break | Name the strip material, its width in mm, and the chamber count |
| High quality hardware | "Quality" is not measurable | Name the operator model, the lock point count and the cycle test report |
| German hardware | A country is not a grade; every brand makes several tiers | Name the brand and the specific product line, plus the report |
| Complies with international standards | No designation, no edition, no class | List each standard with its year and the required class or grade |
| Weatherproof | No pressure, no duration, no definition of failure | Define uncontrolled water, then give the test pressure and method |
| Double glazed | Two pieces of glass is not a specification | Give the full make-up, spacer, coating, gas fill and certification |
| To match existing | Existing what — and measured by whom | Attach a survey of the existing unit with dimensions and photographs |
| As directed by the architect | A direction given after bid is a change, not a requirement | State the requirement now, or state the date by which it will be given |
| Install in accordance with manufacturer's instructions | The instructions were never submitted, so nobody knows what they say | Require the instructions as a submittal, then reference the submitted revision |
| Or approved equal | "Equal" is undefined, so it means whatever is convenient | List the criteria a substitute must document, and the deadline for submitting them |
A specification is only as good as the documents that answer it. Here is the honest map for our own casement range, so you can see which of your clauses we can answer from a catalogue and which have to go back to a test house or a calculation.
| Line | What we publish | What your clause still needs to say |
|---|---|---|
| Alloy and temper | 6060-T66 on every series | Identical across the range, so it is a floor not a differentiator. Ask for a mill certificate per batch, traceable to the packaging |
| Wall thickness | A stated range, for example 1.8 to 2.0 mm | Name the members, state that it is measured on metal excluding coating, and say where it may be measured |
| Thermal break | PA66GF25, multi-chamber; a width band on some series | Require the strip width in mm for the series you are buying, not a generic material name |
| Thermal transmittance | K values on some series, and not on others | These are system values, not whole-window values for your opening. Ask for a calculated value at your size and glass, and name the boundary conditions |
| Glass sash thickness | Figures such as 62, 57 and 81 mm by series | A stack-up, not a glass rebate. Ask for the rebate dimension on the section drawing |
| Frame depth | Figures such as 119, 113 and 116 mm by series | Depth into the wall, not an opening dimension. Confirm it against your wall build-up |
| Mullion | A pair, such as 75/31, 75/21 or 58/22 | First number is depth, second is the visible face. Specify both |
| Corrosion and shear figures | Published for one specific series only | Do not transfer them to another series, and note that neither is an air, water or wind rating |
| Cycle counts | A rig figure for a named hardware brand | State that it is a bench result, and name the configuration you want it demonstrated on |
| Air, water and wind ratings | Not published for the casement series | This is the most important line on the page. A clause naming a class requires a test we must run or a calculation we must produce. Ask before you write it |
That last row is worth restating, because it is where most casement specifications quietly break. A defect is a departure from a declared threshold. Where no threshold is declared, "the window leaks" is not a measurement — it is an argument. If your project needs a rated window, say so in the enquiry and ask what has to be tested, at what size, and to which edition. We would rather be asked in week one than argue in week forty.
One last structural point, and it is the one that costs the most money. The specification lives in the contract documents. The purchase order is what a factory actually receives. They are different documents, and the gap between them is where disputes are manufactured.
We can send product data, section drawings and the test documents we hold. What we will not do is write your section for you, because the two sentences that matter most — the class you need and the size it has to be demonstrated at — depend on your code edition and your elevation, not on our catalogue. Take the skeleton above, fill the blanks, and send it to us with the enquiry.
They do different jobs. A brand gives you a known geometry and a known hardware set; a performance value gives you something you can test. The strongest sections do both: name the basis of design, then list the performance requirements every substitute must document. Neither works alone.
Because load grows with the sash. A small casement and a large one built from the same extrusions do not behave the same way under pressure — the frame, the glass, the hinges, the lock points and the anchorage all carry more. That is exactly why the tested size appears inside the designation rather than beside it, and why certifying bodies require the largest intended size to be the one tested.
It is a start with two holes. Standard-compliance clauses import the entire document, including the classification tables, so you have adopted levels you did not choose. And without a class or grade named, there is nothing to test against. Add the edition year and the required class.
Usually not. Glass is normally specified in the glazing section, and the window section references it. The failure mode is writing a glass make-up in both places and ending up with two answers. Pick one home for it, cross-reference the other, and require conflicts to be raised in writing before fabrication.
The definition of uncontrolled water: water not drained to the exterior, or water appearing on the room side. Most window disputes are not disagreements about products; they are disagreements about what "leaks" means. Define it in Part 1 and half the arguments disappear before they start.