Posted on 28 Comments

Evaluating Combustion Air

Hey, I would really appreciate some feedback on this method of evaluating combustion air.

Unnecessary Holes in the Building

Contractors often cut combustion-air holes in ceilings, floors, walls, and doors without knowing whether the combustion appliance zone (CAZ) needs additional air. These new openings can lead to unintended consequences like pressurizing or depressurizing the CAZ or admitting cold drafts.

Combustion Air Evaluation

Testing an 80+ gas furnace.
Testing an 80+ gas furnace.

The best way to evaluate the combustion air is with an electronic combustion analysis of the building’s combustion appliances. During worst-case testing, the combustion analyzer measures both CO and oxygen (O2). The O2 is an indicator of excess combustion air.

  1. Sample undiluted flue gases as they leave the appliance’s heat exchanger during worst-case conditions.
  2. If the  reading from the combustion analyzer is more than 5% with minimal CO, this indicates that an adequate amount of combustion air is available.
  3. If the O2 reading from the combustion analyzer is between 0% and 5%, this indicates that combustion air is inadequate. We would expect significant CO with this O2 reading


Testing an atmospheric oil boiler.
Testing an atmospheric oil boiler.

At 5% or more of flue-gas oxygen, additional indoor or outdoor combustion air is usually unnecessary. If the O2 is between 0% and 5%, open a nearby window or door a little. If the O2 level rises to an acceptable level, install a combustion air supply from one of these areas.

  • Another indoor space
  • A ventilated intermediate zone, such as a ventilated attic or ventilated crawl space.
  • Outdoors

28 thoughts on “Evaluating Combustion Air

  1. Here is a post from Fred Clade, Tom Andrews’ colleague at the INCAA training center:

    A combustion appliance zone (CAZ) is classified as the space or room containing the combustion appliances. All combustion appliance zones are to be evaluated to determine whether proper combustion and ventilation air is available. Combustion air is supplied to the combustion appliance one of four ways.
    1. To the CAZ directly through air leaks in the building.
    2. To the CAZ through an intentional opening or openings between the CAZ and other indoor areas where air leaks replenish combustion air.
    3. To the CAZ through intentional openings to the outdoors or ventilated intermediate zones like attics or crawl spaces.
    4. Directly from the outdoors to the appliance. Appliances with direct combustion air supply are called direct vent appliances.

    Important Note: The National Fuel Gas Code (NFPA 54 – 2009) presents two methods for calculating combustion air. The simpler of the two methods is The Standard Method. This method may only be applied when air leakage rate of the CAZ or house is sufficient. To use interior air for combustion and ventilation, the estimated natural air infiltration rate of the building must be no less than .4 ACH. If the air leakage rate of the CAZ or structure is determined to be insufficient, then the combustion and ventilation air requirements are to be met using the KAIR (Known Air Infiltration Rate) method per NFPA 54. However, neither method really predicts the amount of available combustion air due to the effects of exhaust fans and pressure imbalances from air handler operation. A comprehensive worst case CAZ depressurization test as described in chapter one must be performed to evaluate the combustion safety of the system.

  2. I’m happy about the level of thought that went into the comments posted here. I learned a lot. Based on the comments, I’ve made another post. Thanks to everyone who contributed. Here is the link to the revised specification.

  3. From Jim Fitzgerald by email: John.
    We all can benefit from better understanding how to find the really dangerous situations for remedies when the budget has limits.

    I think you your interpretation of that test result may be correct. However assuming it applies to other conditions may not be justified. I’m interested if this measurement shows a difference when equipment is overfired but showing little or no CO under normal operations. High CO under downdraft but went away with full clean and tune service that included adjusting input to the label or slightly below.
    With equipment and venting that meets code and mfg requirements and no or infrequent depressurization probably ok. Also CO alarm and spill alarm for alerts as conditions change.

    As far as I know all short term performance measurements can change when conditions change and need to be combined with a comparison of the combustion vent system, the equipment and fuel supply to the fuel gas code and manufacturers’ requirements. “Worst case” testing is an illusion in 20 mph wind at 10F and below, more systems work well.
    Natural draft water heaters can be tough to vent in summer with exhaust & return combining unless you can effectively isolate the appliance and provide the combustion air openings.

    If house and/or duct pressures can affect the vent system and burner, then operations can change.
    If the mechanical room has no openings to the house (see IRC on garage separation from the loving space) and all combustion air from outside, then appliances can perform like direct vented appliances. Combustion air from the attic and crawl space needs adequate venting to the outside and low duct leakage in those spaces.

    Tom Andrews is right about combustion air openings providing makeup air to other equipment with pressure imbalances.

    If a owner wants maximum tightness with no extra openings to outside but can’t watch the details, get all power/direct vent, or super efficient full package with no combustion

  4. From Eric Kjelshus via email: It’s all about the flue. If the O2 is too low just air-seal the return and make a hole in the supply. There are 9 ways to get high CO. If you can just install a sealed-combustion unit then do it! If the flue will not work then don’t use it. If the fan makes the room go negative then fix it. If the boiler, furnace or hot water heater is putting out over 400 PPM in the flue, fix it before you leave or shut it down. Don’t let your job kill or poison others or yourself, do no harm and get paid for it.

  5. If the other levels..CO,Efficiency,stack.on a 90% + are within range, the O2 levels really can be all over the place.. Depends of what your testing. I would drop that level to 3%. Some of the newer variable speed furnaces and modulating boilers I have seen have o2 level below 5% and everything else was within acceptable range.
    Now older 80% if that 02 level is below 5% you will probably have other numbers outta wack.

    1. Rob, good points. I’ll move that O2 number down and provide a few more specifications. One would be to make sure the unit isn’t overfiring before concluding a lack of combustion air. In general, would you say that appliances with more precise draft control have lower minimum O2 values?

  6. John, I believe that over-sizing combustion air vents is deliberate, an attempt by the code writers to alleviate worst-case combustion air problems. Code wording is prescriptive rather than performance based.

    The NFPA publishes a Handbook companion to their National Fuel Gas Code that includes some of their justifications for specific code sections.

    Sometimes, performance testing validates the Code’s prescriptive language, such as when Jim Fitzgerald and, I believe, the Minneapolis Airport Study validated the NFPA-54 Vent Sizing Tables, a few years back.

    Perhaps similar performance validation of Code language might be done for combustion air. Might developing such a validation protocol be a topic for the BPI STC?

    1. I think there is a lot of confusion about combustion air. I think that contractors make holes to the outdoors, which may have unintended consequences, such as pressurization or depressurization. I would support the development of a simpler and more authoritative protocol, you bet.

  7. The 02 reading has nothing to do with combustion air into the CAZ unless the unit has had a CT&E done first. Usually the appliance just needs to be adjusted to let in more excess air. If that fails than maybe this makes sense. I will field test this and see if the info it gathers is useful or not.

    1. Chad, thanks for your comment. My understanding is that 5% O2 represents around 33% excess air, meaning that there is 33% more combustion air than perfect combustion, where O2 would be 0%. Usually either the fuel input is too high or the combustion air is inadequate if the atmospheric appliance is performing at less than 5% with significant CO. Open a door or window and clock the meter and you’ll probably have an diagnosis.

  8. I remember struggling with this several years ago when it was part of my daily job to resolve these type of problems. The “codes” never satisfied me because the numbers seemed arbitrary and inflexible. I could almost always think of a specific situation in which a code would not address a real problem that I might encounter, or, on the other hand, seem to be an overreaction to a need. When I understood the prevailing variables of the appliance and appliance room, I could trust my evaluation of the safety of the situation and walk away with confidence that the appliance would operate safely, regardless of whether or not the code was being demonstrated. My sense is that codes will sometimes prevent someone who doesn’t know what they are doing from making a big mistake, but I would always prefer a situation in which a trained, understanding technician makes a better decision because they do know what they are doing!

    1. Charlie, I agree with your comment. The codes don’t provide perfect solutions and often the guidance is somewhat obsolete.

  9. “In the August 2012 edition of Energy Engineering, Gary Krishman, boiler engineer for the NYC school district, questions the accuracy of the accepted combustion air guidance. He claims that these calculations result in combustion-air vents that may be oversized by two times or more.” – See more at:

  10. David Richardson from Comfort Institute wrote on LinkedIn: There are other issues that can cause O2 levels to be below 5%. I would be concerned that someone who encounters readings as you’re describing might start chasing a combustion air problem that doesn’t exist. There are additional readings needed to differentiate between what could be a combustion air issue or an issue with the firing rate of the equipment. Draft pressure and stack temperature are two of those readings needed to make this distinction.

    1. David, I agree with your comment which to me means, “Learn and understand what the analyzer and other testing tools tell you before doing anything.” I would add, “Base your decisions on the testing rather than some guidance that predates our current testing protocols.”

    2. True,

  11. I’m concerned about relying on combustion testing to evaluate combustion air. So many things can go wrong during the test. Plus this is a spot test. I think that the intent of Code combustion air requirements goes beyond building science-type worst-case test conditions.
    I suggest that after any combustion air opening are made to indoor conditioned space that they first meet code, and pass a BPI-compliant post-installation worst-case draft and combustion test. And that when consider installing one or two combustion air vents to a ventilated attic, make certain there is enough attic ventilation to both ventilate the attic and supply combustion air/ventilation air to the CAZ; plus be aware of the common wind conditions in the area: high wind can depressurize both the attic and the CAZ; and certainly run worst-case testing afterwards. When doing a high vent into a ventilated attic and a low vent into a ventilated crawlspace, ensure the vents to the outdoors are permanently open and adequately sized, that the floor and ceiling vents are equally-sized, and do a worst-case afterwards.
    I’ve been thinking about this a bit lately: Systems Thinking. In some ways, good HVAC professionals have a hand up on Systems Thinking over Weatherization pro’s, even those who are steeped in Building Science. HVAC is an often-complex system, and a good contractor has to know how to size the mechanicals, the distribution system, the combustion air, make-up air, and house ventilation air; how to eliminate cold spots in the house and to evenly distribute conditioned air throughout the house; and ensure and maintain both comfort and safety, yet do so in an affordable manner. This is systems
    But where many traditional systems thinkers in the HVAC community fail, is in their understanding of the growingly complex interactions between HVAC systems and Building effects: thermal bypasses, moisture sources and transport, zonal pressure differences, stack effects and wind effects, under-ventilated and over-ventilated unconditioned spaces, infiltration control and unexpected mechanical de-pressurization. These effects are outside the box from traditional HVAC systems thinking, and unless an HVAC contractor thinking. either works directly with a Weatherization contractor or adds weatherization contracting into their services, addressing Building system problems will cut into their bottom line.
    Thus it is left to people like John Krigger and the folks at BPI to find new ways to translate the analysis of complex HVAC system-Building effect interactions into practical and repeatable diagnostics that can be used by HVAC professionals cost-effectively and yet meet at some level the rigors demanded by Building Scientists. Such new diagnostics require public discussion and ultimately technical committee work, as well as approval by code-writers.
    John, thanks for formulating this combustion air evaluation protocol. I do not think it will fly with the code writers, Perhaps you you could add a proviso that code requirements for combustion air, ventilation air and make-up air be met fiest, and then use your test protocol to evaluate the effectiveness of the installation.

    1. Dave, I’m glad you ventured into the realm of philosophy because it gives me an excuse to talk about the philosophical aspects of this discussion. The amount of safety testing and analysis should be directly related to risk. Over-testing, over-calculating, and over-thinking safety issues wastes resources without necessarily resulting in better safety compared to a simpler approach. The greater the complexity the lower the compliance.

      Yes, I’m out on a limb with this post. One could ask, “Where is your proof for this approach.” I’d respond, “Where is NFPA’s proof that their calculations and prescriptive measures result in safety without testing, which in my 2009 version of NFPA 54 isn’t mentioned in the combustion air section?” Author and risk expert, Doug Hubbard, argues that perfect information is too expensive. You can assert that it’s worth millions to save one life, but what if you could save a million lives with the same resources? The closer you get to perfect information the more unreasonable the cost and you can’t bring the risk to zero anyway.

  12. As much as I agree with testing the appliance to see if it has enough combustion air…code in out area requires combustion air ventilation unless the unit is “sealed combustion”

    Until the code changes we have to cut the holes

  13. I really do find this evaluation process interesting. My initial reaction (I need more time to process it fully) is that it addresses 1/2 of the conversation. The code section that covers this is titled “Air for Combustion and Ventilation”. This combustion analyzer process may, at times, address the combustion air side but there is no way to validate the ventilation aspect of the space. If there is a possibility of flue products venting into the CAZ then it is difficult to separate these two in the conversation. You would still need to attach your eyes and a tape measure to the space.

    I need to comment on make-up air. As an example, I have been in many homes that have sufficient volume and leakage rates to easily provide the code combustion/ventilation requirements. Combustion readings are fine on the 80% furnace until you turn on exhaust fans. Now the 02 reading drops from fairly normal to under 5%. It may still meet minimum draft pressure and CO numbers but there is a problem. Opening a window will make it work again but this is not a combustion air problem. This is a make-up air problem. In this type of scenario, the holes to the exterior do not follow combustion air rules anymore – sizing protocols can be ignored. Providing a passive make-up air opening to the outside to fix this is going to have to be large enough to accommodate the cfm of the exhaust at a much lower delta P (whatever the CAZ pressure is)than the exhaust fans are providing to dump it outside. The hole is very likely to be prohibitively large.

    I agree that there are a fair number of people out there putting holes in buildings shouldn’t be there and they are being put there for the wrong reason.

    I am always in favor of testing to make sure we don’t miss something. Combustion analysis can certainly tell us things we couldn’t see otherwise. I need time to digest the 5% number. Design 02 numbers vary a fair amount anymore – not like it used to be with a 50% excess air number for most every furnace out there. There is no indication of what types of appliances are being tested in this post. My feeling is that if I had a reading below 5%, you could certainly open a window and see if it gets better. At that point, make the determination if it is a combustion air issue or a make-up air problem. If the number doesn’t change, there are certainly other issues that can cause a low 02 number – dirty burner, dirty heat exchanger, air mix, firing rate…etc.

    1. Tom, you make some very relevant points here. The code talks about combustion air, dilution air, ventilation air, and make-up air. Isn’t this all the same air? Using your quadrant approach during worst-case testing, aren’t we evaluating the overall air supply and identifying the specific problem?

      1. I would argue that, no, it is not all the same air even if it is all coming from the same place. Or, at least does not manifest itself the same way. Combustion and dilution air (if needed) support the combustion process and venting. The ventilation aspect comes into play if there is a malfunction. If the appliance decides to vent into the house for some reason, the space has to be big enough and leaky enough to dilute the flue products down to the point that they (supposedly) will not kill you (this is an entirely different conversation). You are, sooner or later, supposed to say to yourself “hmm, do you smell something?” or “I wonder where all this water is coming from?”.

        Make-up air for exhaust fans is the most difficult issue and, as a rule, we do not install “combustion air” openings to fix a make-up air problem. Section states “Where exhaust fans, clothes dryers and kitchen ventilation systems interfere with the operation of appliances, make-up air shall be provided”. Not a whole lot of guidance on how to do it though. Here is another example of this make-up air issue. Section states “Make-up air shall be provided for Type 1 clothes dryers in accordance with the manufacturers installation instructions”. If you look at older code books, this section used to say something along the lines of “Type 1 clothes dryers must be vented to the exterior and no other fuel burning appliance can be installed in the same space”. This was because of the depressurization problem with dryers back-drafting combustion appliances installed in utility rooms. This requirement was ignored and ignored until they just removed it from the code and in came the requirement for make-up air being provided. Again, no guidance on how to do it though.

        To put a passive air opening into a building to provide make-up air can easily be prohibitive due to the size required. Let’s say you had a 1500 cfm50 house that was large enough and leaky enough to provide combustion and ventilation and the appliances work just fine. But, during worst case, the house is a -3 pa with a total of 250 cfm of exhaust. (not unreasonable). The water heater doesn’t work and you open a window and now it works. Look at the size of the “combustion air” opening you would need to add. I calculate it would take a 14″ round pipe to provide the 250 cfm make-up air needed at a 3 pa pressure diff. Although, you could just put in enough to make it work and leave the appliance at the edge of the cliff.

        The quadrant approach is intended to identify depressurization problems but there is no one specific solution for addressing it. The chart identifies if the problem is ducts, doors or exhaust. Combustion air is addressed in a different manner from make-up air problems or duct/door problems. Combustion air as a solution is not on the quadrant chart – that should have been addressed earlier in the process and verified during combustion testing.

        John, I am sorry this is so long.

      2. John, I am really sorry this is so long.

        I would argue that it is not all the same air even though the source is the same. Or, at least, they manifest themselves in different ways. Combustion and dilution are used in the combustion process for proper combustion and venting. Ventilation is of concern when there is a problem. Assume a draft-hood equipped boiler has a plugged vent. The intent is that the flue products need to vent to the outside or into a space that is big enough and leaky enough to dilute the flue products down to the point that they don’t kill somebody (or, at least until you say “what is that smell” or “where did all this water come from?” (whole ‘nother conversation here).

        Make-up air is the problem. Section of the 54 states: “Where exhaust fans, clothes dryers, and kitchen ventilation systems interfere with the operation of appliances, make-up air shall be provided.” Section states: “Make-up air shall be provided for Type 1 clothes dryers in accordance with the manufacturers installation instructions.” Now this is interesting in that older code books said something along these lines: “Type 1 clothes dryers must be vented to the outside and no other fuel burning appliances shall be installed in the same space.” This was because of the depressurization from the dryer back-drafting combustion appliances. This provision was ignored to the point that they just eventually gave in and changed the code. My belief is that combustion air and make-up air are two separate and distinct things and are treated differently. There are plenty of spaces that are large and leaky enough to adequately provide, per code, combustion and ventilation for appliances – UNTIL – we do worst case and create depressurization from exhaust fans.

        Providing make-up air follows a different set of rules than combustion air openings. As I alluded to earlier regarding sizing, imagine a 1500 cfm50 home that meets combustion and ventilation requirements and the appliances work just fine. Turning on exhaust (assume a quite reasonable 250 cfm total exhaust) depressurizes the home to -3 pa and now the water heater malfunctions. By my calculations, it would take a 14” round “combustion air” opening to the outside to provide the 250 cfm make-up air for the exhaust. Code combustion air sizing guidelines are inadequate. There is no code guidance for providing make-up air.

        I would say that the quadrant isn’t necessarily used for “evaluating the overall air supply”. It is intended to identify causes and solutions for depressurization problems. It deals with ducts, doors and exhaust. Combustion air is not mentioned as a solution for the problems. That is something that is addressed earlier in the process and is evaluated during combustion testing.

        1. Hey don’t worry about the length of your post; I wouldn’t have written this blog unless I was searching for answers. Wouldn’t it be great if make-upair was designed into buildings. Or better yet, sealed combustion and balanced ventilation.

          Lacking the time, money, and skills to diagnose and fix these air-related problems, is worst-case testing with combustion analysis enough to reduce risk to an acceptable level?

  14. Regarding the submission by Jonathan, I would be interested in knowing whether it was the furnace or the water heater that was the issue and whether the testing done both times was under worst case.

    I would argue that the problem with the CAZ may be a make-up air problem, not a combustion air issue. Pressure relieving the utility room to the rest of the house now allows the dryer to suck on something other than just the CAZ. Certainly the room is lacking the proper volume so both issues were addressed with the addition of an opening to the rest of the house.

    But, it would not surprise me to see the furnace function just fine if it was tested under normal conditions without the dryer operating – even with the door closed without the new opening.

    The other thing that seems to be forgotten in these conversations is the ventilation aspect. Just putting one opening into the space (particularly low) would be inadequate to ventilate the space in the event one of the appliances decided to vent into the CAZ for some reason. Code requires openings high and low.

    1. Tom, thanks for your thoughts on this. The code was written before we started doing worst-case testing. The code is a prescriptive approach. Is our current worst-case procedure and consideration of O2 a better approach?

      1. I used to have the NFPA 54 codebooks back to ’72 when they first started putting them out in that format. “Worst case” testing was a part of the Appendix in all of them. This is not something new. Carrier Corp even did a study back in the ‘70’s looking at what time of day worst case was likely to occur. I don’t have an issue really with the prescriptive approach as I can only imagine where we would be without it. Certainly times have changed and I concur with your inference that buildings and equipment have evolved beyond the code. Yes, I agree that current testing procedures and consideration of 02 is a better approach.

  15. Submitted on Linkedin by Jonathan Finewood: I recently did an audit where gas furnace, dhw, and dryer were in a confined space. O2 was 3% and CO was 900ppm during worst case!! I called the gas company to inspect and red tag it. When they got there he tested with his sensit. He found CO was around 50ppm. I was like wait a minute….how did I get 900 about an hour ago and he got 50 just now?? I noticed the door to the CAZ was wide open and shut it. He tested and CO shot back up! We both came to the conclusion that there was not enough combustion air and recommended to cut some in a wall adjacent to the space. At test out, and a clean and tune, we were good! O2 was at 7% and CO was down to 35ppm.

Leave a Reply

Your email address will not be published. Required fields are marked *