What Is the NFRC Rating Process for an Aluminum System Windows Manufacturer: Nobody Certifies Your Window, Including NFRC

The Answer, and the Sentence Inside NFRC's Own Document That Nobody Quotes

Here is the short version. An aluminum system windows manufacturer does not send a window to NFRC and receive a rating back. The manufacturer joins a program, hires two independent parties, builds one physical sample, and then performs the certifying act itself. NFRC's own published procedure document says this in plain language: "NFRC does not certify a product and certification does not constitute a warranty of NFRC regarding any characteristic of a fenestration or fenestration-related attachment product." The same disclaimer continues: "Certification is not an endorsement of or recommendation for any product or product line or any attribute of a product or product line."

So what does NFRC actually do? It licenses and administers a process in which three independent parties each perform a function the other two cannot. NFRC runs the rating system and keeps the directory. An accredited laboratory models and measures. A licensed inspection agency reviews the paperwork, walks the factory floor, and recommends certification. The manufacturer then labels the product, and that labelling is, in NFRC's own definition, the certification.

There is an honest tension worth naming, because it tells you more than a clean explanation would. NFRC's public product certification page says at one point "we certify the product." The legal disclaimer inside the program document says NFRC does not certify a product. Both are published by the same organization. We are not going to adjudicate which sentence is more correct. We are going to point out what the coexistence of the two sentences means for you: an NFRC rating is a manufacturer's declaration that has been made checkable by other people's work. The check is real. The declaration is still the manufacturer's.

Everything below follows from that one idea, and it is written for the specific case in the title: a company that makes aluminum system windows. That phrase bundles together two things NFRC treats as belonging to different parties, and section eight is where that becomes expensive.

Five Parties, Four of Them Independent, and Only One of Them Signs

The program only makes sense once you can name who does what. Note the pattern in the last column: four of the five parties are structurally unable to do the thing buyers assume they do.

Two details from the current agreement structure are worth pulling out, because they are the ones that catch exporters. First, the entity granted certification is called the Certification Holder, and it is authorized to label under its own three-letter or numeric manufacturer code and product codes in the directory. Second, that same document requires a license agreement from any legal entity separate from the Certification Holder which labels, manufactures, sells, or distributes products under the holder's code, and it names fabricators, private labelers, pre-hangers, distributors and sellers explicitly. In other words, the name in the directory and the building that made the window are allowed to be different legal entities. Both sign.

When NFRC revised the license agreement in 2025, its own frequently-asked-questions page explained why it now asks for more on Schedule I. It needs to know "where certified products are being manufactured, whether that manufacturer is affiliated with the licensee, and whether licensee's certified products are being labeled and distributed under company and product names different than what is reflected in the certified product directory." Read that twice. It is the program administrator telling you that the gap between the listed name and the actual building is a known failure mode, and that it has now written a question specifically to catch it.

What Is Actually Being Rated: A Product Line, Not a Window

You do not submit a window for a rating. You submit a product line. The program document is specific about what the licensee hands to the simulation laboratory: a bill of materials, product assembly drawings, detailed dimensioned die drawings, and all individual product options within the product line. The laboratory is then directed to simulate each individual product and develop a matrix.

The number on a label is one cell in that matrix. Your window has to be one of the cells. If your configuration is not in the matrix, it is not rated, no matter how closely the series name matches. This single fact is where most disputes come from, and it is the first thing any experienced buyer checks. A sourcing consultancy that works with Chinese factories publishes a five-check list for exactly this problem, and its second check is: confirm the series, the operator type, the glass make-up (glass thickness, low-E product and surface number, gas fill, spacer type) and the frame colour and finish are all inside the certified options. Its own words on the glass: "A change of spacer or a swap of low-E product moves the number and can invalidate the label."

The program also anticipates that you will want to change something later. The product requirements document states that before adopting a modification in the design or construction of an authorized product line, the change must be submitted through the waiver application. A modification that is still within the same product line requires submitting performance showing it is permitted. A modification that requires a separate product line is processed as a brand-new product line, following all the rules from the beginning. There is no third category where a change is simply a documentation ｕｐｄａｔｅ.

The Validated Computational Procedure: One Physical Test Anchors an Entire Matrix

Most people picture a window sitting in a hot box. For most ratings, that is not what happens. There are three evaluation routes in the program, and which one applies depends on which rating you are chasing.

The validation step deserves a closer look, because it is the only place where a real window and a real rating touch each other. The licensee manufactures one representative sample, delivers it to an accredited testing laboratory together with drawings, the bill of materials and the program's submittal form, and has it tested to the thermal test procedure. The laboratory sends the report to the inspection agency. The matrix is considered validated if the physical result meets the equivalence criteria in the rating procedure.

Two further wrinkles. If your product line sits inside a Validation Test Matrix, one validation test validates every product line in that matrix, which is the cheapest thing in the whole program. And the samples submitted for initial certification serve two purposes at once: validation of the baseline product and plant qualification. One box, two clocks, and the second clock is the one people forget.

Air Leakage Is the Only Rating That Must Be Measured, and It Is Optional

Set the two facts side by side and they sound like a mistake. The solar heat gain and visible transmittance procedure states that U-factor, solar heat gain coefficient and visible transmittance are the mandatory ratings for labelling, and that air leakage and condensation resistance are voluntary supplementary ratings. The product certification program states that air leakage ratings can only be obtained through physical testing, because "there are no procedures for the simulation or computation of air leakage ratings."

So the one rating on the label that is always a measurement is the one rating a manufacturer may decline to publish. And the rating everybody quotes, U-factor, is a simulation validated by a single physical unit. Neither fact makes the system weak. Together they mean something practical: when you ask for a U-factor you are asking for a calculation, and when you ask for air leakage you are asking for a measurement. Specifiers routinely ask for the first with great precision and treat the second as an afterthought. That is backwards if the building is in a windy, cold climate, because infiltration is the part that depends on how the unit was assembled in your plant that week, not on how the profile was drawn.

There is a second mismatch. NFRC's own explanation of the label shows air leakage at the top of its scale, and at least one major energy code requires the temporary label to certify compliance with air leakage requirements. That means a rating which is optional for the manufacturer is mandatory for the code in that jurisdiction. Before you write a specification that only asks for U-factor, find out whether your authority treats air leakage as an optional extra or as a condition.

The Model Size: NFRC Rates at a Size Nobody Orders

This is the section most buyers need and few have read. Ratings are produced at fixed model sizes defined by product type, not at the size on your drawings. The table below reproduces the size grid as published in the protocols manual of an adopting national rating council, which quotes it from the rating procedure and notes that it is expressed height by width.

Nobody has ever been shy about this. Reported ballot language in the trade press put it in one line: "NFRC rates products at a standard size and should not be confused with actual product performance." The same reported committee exchange contains a better line still, from a manufacturer's representative objecting to a proposed change to the casement sizes: the current sizes were selected to represent common casement products, and changing them "would require us to build something for simulation that nobody in practice actually manufactures." That is the cost of the grid, stated by the people paying it.

The adopting council that publishes the grid is candid about its own mismatch too, writing that "it is recognised that the standard simulation configurations are not representative of the typical window configurations found in Australian housing." A national body adopted the method and then wrote down that the method does not describe its market. That is worth more than any marketing claim about how universal a rating is.

Now the arithmetic, and this part is ours, with the inputs shown so you can substitute your own. Take a 48 x 48 casement window. At 25.4 mm to the inch, that nominal size is 1219 by 1219 mm.

In an aluminum frame, the frame is where the heat goes. So a unit with 1.4 times as much frame per square metre of glass is not the same thermal object as your window, and the label number was produced on the frame-heavy one. For a square window like this, the direction of the difference is probably in your favour. For a very large unit, it may not be. What is safe to say, and what nobody can argue with, is that the rated object and the purchased object are two different windows. Codes commonly accept the labelled figure. Your energy model may want size-adjusted values, and a good consultant will ask for them.

The grid also separates frame materials by product line rather than by physics. A vinyl casement window and an aluminum casement window are rated at the same model rectangle and are still two entirely separate product lines, each with its own simulation file, its own matrix and its own listing. And casement window vs double hung is not a styling preference inside this system: the grid assigns the two types different model sizes, so they are different specimens from the first line of the file.

The Sightline Rule: A Small Change Re-opens the Whole File

At a recorded committee meeting reported in the trade press, a manufacturer's representative described the then-current grouping rules this way: "The way the wording is in the grouping section, if you make any change to sightlines whatsoever you have to re-simulate." The increments under discussion were as small as one sixteenth of an inch.

One sixteenth of an inch is about 1.59 mm. Two of our own outward-opening series differ in frame width by 3 mm, which is roughly twice that increment. If that reading of the grouping rule holds, then a 3 mm sightline difference between two series is not a cosmetic variant. It is a new simulation file.

The practical instruction is blunt: freeze the sightline, the spacer, the low-E product and its surface number, the gas fill and the frame colour before the simulation file is built. Changing any of them afterwards is not a documentation task. It is either a waiver submission with supporting performance, or a new product line that starts the process again.

The Aluminum System Problem: Lineal Supplier and Fabricator Are Two Licensees

This is the section that exists because of the word system in the title. An aluminum system window is the combination of two things that the program treats as belonging to different people: the lineal, meaning the extruded profiles and often the thermal break strip and hardware platform, and the fabricated window that someone assembles from it.

The program has separate licensee categories for exactly this split, including Manufacturer, Fabricator, Door Manufacturer or Lineal Supplier, and Private Labeler. And it contains one mechanism that governs how work moves between them. On simulation: "A Lineal Supplier shall authorize each Fabricator to obtain the Lineal Supplier's simulation report by authorizing the simulation laboratory to reissue the report to the Fabricator." The report must include all individual product options authorized for certification. The same mechanism is repeated for the test report.

Read that carefully, because it is the whole procurement risk in one sentence. The simulation does not travel by itself. It does not travel because the profiles came from the same supplier, or because the fabricator is a licensee of the system house, or because the brochure says the system is certified. It travels only when the Lineal Supplier instructs the laboratory to reissue it, in writing, to a named Fabricator. "Our system has been simulated" and "your fabricator holds a certified product line" are two different sentences, and one written authorization is the only thing that turns the first into the second.

If you have read anything about European compliance, this shape will look familiar. Under the European construction products framework, a manufacturer may reach a declaration by using another manufacturer's test data with written authorization, or by using a system supplier's data under a cascading arrangement. Here the instrument is different and the idea is identical: a test performed on someone else's assembly becomes yours only through a documented act, and that act is performed by a laboratory on someone else's instruction.

The Plant Is Inside the Certificate

A rating is not attached to a brand and it is not attached to a catalogue. It is attached to named buildings. That is what the schedules in the license agreement are for.

Two separate clocks then run against each of those facilities. The first is plant qualification. Each plant or fabricator location listed on Schedule I or III must test at least one production line unit every five years, and the result must meet the same equivalence criteria as the original validation. A plant added partway through a cycle must submit a unit within six months of being added. If there are more product lines than plants, each plant tests a different line; if there are more plants than lines, the same line is tested again.

The second clock is inspection. The product requirements document states that not less than once in each year the licensee's designated inspection agency shall conduct an inspection of each production facility. The inspection agencies' own operations manual sets the first of those at not less than six months after initial certification, and annually thereafter, with up to forty-eight hours' notice permitted. Findings are reported in writing within fifteen days, the licensee has fifteen business days to correct, and failure to correct can lead to suspension. A licensee that refuses access to a facility without good cause gives the program grounds to deny, withdraw, suspend or terminate certification on that basis alone.

On top of both clocks sits a quality management system requirement: designated competent personnel, supervision of production at each facility, and regular audits of authorized products so that there is correlation between the fully assembled product and what is described on the label. The program is not checking your window. It is checking whether your factory can keep making the window it was rated for.

Your Insulating Glass Carries Its Own Certificate, and You Do Not Own It

The product requirements document puts this duty on the licensee in one sentence: licensees shall be responsible for assuring that all insulating glass units used in authorized fenestration products are certified in accordance with a participating insulating glass certification program. The current license agreement repeats it as a licensee function. Schedule I now asks which program each facility participates in.

Two consequences follow. First, your window rating is contingent on a certificate held by a glass supplier, issued by a programme you do not control and cannot renew. Second, the program has already built the contingency rules: conditional insulating glass certification is accepted for a period of up to nine months, and if a participating glass programme is delisted, NFRC notifies licensees within thirty business days and provides a transition plan.

The commonly referenced names in this part of the world are the insulating glass programmes administered by the same service organisation, whose own frequently-asked-questions page also notes that its window programme audits are annual while its insulating glass audits are twice a year. That asymmetry is the whole point: the thing inside your window is audited more often than your window is. If your glass supplier's status lapses, the thing that lapses is not a document about glass.

Residential, Commercial, Component Modelling, Site-Built: Four Different End Products

Buyers often ask for "the NFRC certificate" as though there is one artefact. There are at least four, and they belong to different things.

The symmetry worth carrying away: a residential label belongs to a product line and follows the unit; a commercial label certificate belongs to a project and follows the address. NFRC's own site says label certificates are used in commercial buildings, are project-specific, and include the address and contact information for the project as well as the certified ratings. If you are buying for one building, you are most likely buying the second thing, and it is not the same document as the first.

The Label Is Part of the Rating

The last step in the process is the one that most often fails at the worst possible moment, because it happens in a packing bay rather than a laboratory.

A sourcing consultancy that works with Chinese factories states the failure mode without softening it: if units arrive unlabelled, the site inspector has nothing to accept, and re-labelling on site is not permitted. Treat that as a factory operation with a sequence, exactly like handing. It has to appear in the order documents with a named responsible party. If the labels are applied after packing, or applied by a trading company at a warehouse, then the certifying act has moved away from the plant that was inspected, and the schedule that records where labels are applied is the document that will show it.

What the Rating Does Not Cover

An energy rating is an energy rating. It is not a structural rating, a water rating, a security rating or a fire rating, and the directory listing will not tell you that unless you already know.

The same sourcing consultancy puts the consequence in one sentence that we would not improve on: "A window can be excellent on the NFRC label and unrated structurally, or the reverse." Note also the direction of the queue. NFRC describes itself as the official certification body for the ENERGY STAR programme for windows, doors and skylights, and states that products must be NFRC certified before applying for ENERGY STAR certification through its own FenStar pathway. The sequence runs from the rating outward, never the other way.

What an Unrated Product Falls Back To

There is a legal route for a window with no rating, and it is worth knowing because it is the alternative against which the cost of the process should be measured. In California, the energy code designates NFRC as the supervisory entity responsible for administering the state's programme, and it permits a product that has not been rated by NFRC to carry a temporary label bearing the words "CEC Default U-factor" and "CEC Default SHGC" followed by the value from the published default table.

That published table includes entries for metal frames and for thermally broken metal frames. In the units the table uses, the values are 1.28 for metal operable and 1.19 for metal fixed, against 0.66 for thermally broken metal operable and 0.55 for thermally broken metal fixed.

Now our arithmetic on those published numbers, given here only to show the size of the penalty. The gap between metal operable and thermally broken metal operable is 0.62, and between the two fixed entries 0.64. In other words, on a thermally broken aluminum window, not having a rating roughly doubles the U-factor you are allowed to claim. That is an expensive way to avoid a process fee, and in any competitive specification it loses the job outright. The default route exists so that a project can be permitted, not so that a product can win.

The Sequence, Step by Step, for an Aluminum System Windows Manufacturer

Everything above, in the order it actually happens. The middle column is the part that is usually discovered too late.

Where Our Own Published Numbers Stop

We are the factory writing this, so the useful thing we can do is say precisely which of our numbers survive contact with this process and which do not.

Three more things, and these are the ones we would rather you heard from us than discovered later.

We are not listed in the NFRC Certified Products Directory. We are not going to imply otherwise by quoting a laboratory report as though it were a certification. If your project requires an NFRC rating, ask us directly and we will tell you what we can document, what we would have to build, and what we would have to become a licensee to claim.

We have two production centres, in Foshan and Zhaoqing. In the terms of this programme that is not one factory with two addresses. It is two entries on Schedule I, each separately qualified with its own production line unit and each separately inspected every year. Any certification conversation has to state which one will make your order, in writing, before the first unit is cut. You can see how we handle multi-plant work on our project pages.

Our sliding window catalogue contains one product, and it is a side-press window. It is not a conventional two-track slider. The model-size grid in the rating procedure has a line for a horizontal slider; it does not have a line for a side-press mechanism as such. If you are specifying that product for a North American project, the operator type is one of the first questions that has to be answered, and we would rather raise it in an article than have it discovered at the simulation stage.

What We Can Give You Against What You Must Obtain

Twelve Questions That Separate a Real Listing from a Brochure Claim

Ask these of any supplier, including us. The third column is the one that matters, because those answers are all true, all polite, and all empty.

When Not to Pursue a Rating

Questions We Are Asked Most Often

Does NFRC certify my window?

By its own published disclaimer, no. NFRC administers the programme, accredits the laboratories, licenses the inspection agencies and maintains the directory. The certifying act is performed by the manufacturer or responsible party when it labels a product whose ratings were reported by accredited laboratories and authorized by a licensed agency. Read practically, this means the rating is a manufacturer's declaration that three other parties have checked.

Can a factory outside North America hold a rating?

The programme does not require a North American address, and its own forms include a country field for facilities outside the United States. What it does require is that each facility be named on a schedule, qualified by a production line test, and inspected by a licensed agency, initially and then every year. That means an agency representative travels to the plant. The constraint is logistics and cost, not eligibility.

If the system house has a simulation, is that enough for the fabricator?

No. A lineal supplier must authorize each fabricator to obtain its simulation report by instructing the simulation laboratory to reissue it, and separately for the test report. Without that reissue, the fabricator has nothing to submit to an inspection agency. This is the single most common misunderstanding in system window supply, and it is why "our system is certified" is not an answer to "is your window listed".

Why is the number on the label not the number for my window?

Two reasons, and both are stated by the programme itself. First, ratings are produced at fixed model sizes by product type, and the programme's own reported language says rates at a standard size should not be confused with actual product performance. Second, the label value is one cell in a simulated matrix validated by one physical unit, so a configuration outside the certified options is not covered at all. Codes commonly accept the labelled figure; your energy model may need size-adjusted values.

Is ENERGY STAR the same thing?

No, and the order matters. NFRC describes itself as the official certification body for the ENERGY STAR programme for windows, doors and skylights, and states that a product must be NFRC certified before it can apply for ENERGY STAR certification, through its own FenStar pathway. An ENERGY STAR claim without a rating underneath it is not a claim you can verify in the directory.

What is the cheapest thing to get right before starting?

The bill of materials and the option list, frozen. The simulation laboratory models every individual product in the line, so anything you intend to sell has to be in the file at the start. Adding a spacer, a coating or a sightline afterwards is not paperwork. The programme also requires that any modification be submitted through the waiver process before it is adopted, and a modification that falls outside the line is processed as a new product line from the beginning.

The Short Version

• NFRC does not certify your window. The manufacturer does, by labelling it, after three independent parties have each done a job the others cannot.
• What gets rated is a product line, in the form of a simulated matrix, anchored by one physical test of one baseline unit.
• U-factor, solar heat gain and visible transmittance are mandatory and mostly simulated. Air leakage is always measured and is voluntary. That inversion surprises everyone.
• Ratings are produced at fixed model sizes. The rated object and the purchased object are two different windows, and the programme says so itself.
• Small changes re-open the file. A sightline change of a fraction of a millimetre has been described as requiring re-simulation.
• For aluminum system windows, the lineal supplier and the fabricator are different licensees. The simulation only travels between them when the supplier instructs the laboratory to reissue it.
• The plant is inside the certificate: named on a schedule, qualified by a production line test, inspected every year.
• Your insulating glass carries its own certificate, held by your supplier, under a programme you do not control.
• A residential label follows the unit. A commercial label certificate follows the project address. They are not the same document.
• If the temporary label is not on the unit when it arrives, there is nothing for the inspector to accept, and re-labelling on site is not permitted.