Within Tolerance - Reviewing the TAB Report
Updated: 19 hours ago
Alex Mathers, P.E., CxA, NEBB CP
In theory, a test and balance (TAB) report proves that the HVAC systems deliver design intent. If every fan, pump, terminal and outlet reads within tolerance of design, then every room gets its air and every coil gets its water. In practice, a report can pass at every fan and every box total while an outlet delivers half its design airflow, a pump runs 50 percent over its scheduled flow and a lab’s required ACH is not met.
The commissioning agent's TAB review exists to find that gap. This post draws on our recent reviews of a university laboratory building and a four-building university campus. On the laboratory building, total exhaust measured 83,726 CFM against 85,120 CFM design, and 52 of 53 VAV boxes landed within tolerance. On a summary read one may accept the report and move on. We raised fourteen items, seven of them non-conformances that had to be resolved before acceptance.
Most of what we found came from gaps between the drawings, the TAB sheets and the controls rather than from poor balancing.
The acceptance basis
The review starts with the governing specification because the acceptance basis changes from project to project. On the laboratory building, the criteria came from project sections written by the design engineer: 23 0594 for water systems, 23 0595 for air systems and 23 0598 for rotating equipment, with a tolerance of plus or minus 10 percent. Section 23 0594 also prohibits an incomplete report. Because a floor’s heating hot water system never reached design flow, the balancer issued a partial report, which made the report a non-conformance before it was submitted.
The campus specification set tighter outlet limits, plus or minus 5 percent above 200 CFM and plus or minus 10 percent at or below 200 CFM, and it allowed no negative deviation on total airflow.
The field notes
The balancer's embedded field notes often diagnose a problem before the reviewer does. On the laboratory building, the notes explained why the floor never reached design flow. One reheat valve was cross-piped and read a negative differential pressure. Three more reheat valves on the same floor could not establish flow, with the balancer noting insufficient differential pressure across the flow control valve and a possible dirty strainer or airlock. The chilled water shut-off head test had also been skipped at the mechanical contractor's request, which left the pump curve unverified.

Outlets and box totals
A passing box total does not show that each room receives its design airflow, because high and low outlets cancel inside the box. On one campus building, all 35 terminal box totals passed and the building-wide terminal total was within 1.3 percent of design. Easy, simple, done? Not so fast. Of the 113 measured outlets, 28 (25 percent) were outside the specified tolerance.
As an illustration, a box serving two 200 CFM diffusers that read 230 CFM and 170 CFM delivers exactly 400 CFM and passes, while each diffuser sits 15 percent off design and fails. The worst campus outlet read 100 CFM against 195 CFM design, 49 percent low, which usually points to a closed or blocked branch damper rather than a balancing adjustment.
How the number was hit
Speed, static pressure and brake horsepower show how a system reached its design flow, which the flow reading alone cannot. On the laboratory building, the chilled water pumps moved 675 GPM against a scheduled 450 GPM, about 50 percent over, at 60 Hz with 40 ft of head against a 60 ft design. One air handler reached design airflow only by running the supply fan VFD at 73 Hz. Under the fan affinity laws, power varies with the cube of speed: (73 / 60)3 = 1.80. On the same system curve, the fan draws about 80 percent more power at 73 Hz than at 60 Hz, which indicates more static pressure than the design expected.
The campus fans showed the same pattern:
One supply fan array measured 3.57 in. w.g. external static against 2.50 in. w.g. design, 43 percent high, with total static below design. That combination points to a duct-side restriction rather than a fan capacity problem.
One exhaust fan ran at 65 Hz and drew about 2.6 times its design brake horsepower while moving 4.8 percent less air than scheduled. An inlet static of 1.37 in. w.g. suggested a restriction upstream.
A fan coil was 8.4 percent short on airflow with low external static at a 20 percent speed setting, which is a fan speed fix rather than a duct problem.
In each case, the speed, static pressure or brake horsepower identified a cause that the airflow reading alone could not. On the supply and exhaust fans, that cause also carries an energy cost the owner pays for every hour the fans run.
The design baseline
Each terminal carries three airflow values that should match: the scheduled design airflow, the design value on the TAB sheet, and the minimum and maximum setpoints in the building automation system (BAS). On the campus project they often did not:
On one building, the TAB sheet design disagreed with the current mechanical schedule on 11 of 35 terminals. Two boxes appear to have had their values swapped, 3,050 CFM and 3,250 CFM. One minimum was balanced to 210 CFM against a scheduled 125 CFM.
On another building, the balancer noted “Design CFM doesn’t match floor plan” on 27 boxes.
An RFI response revised the design airflow on 49 VAV boxes across two buildings. Each one then needed a BAS setpoint update and a TAB re-verification.
During the warranty period, trend data showed a box with a BAS maximum of 650 CFM against a scheduled and balanced 470 CFM. Another box was balanced and programmed at 825 CFM while the schedule still showed 1,100 CFM.
The laboratory building had the same gap. Several lab exhaust sheets carried floor plan design values that did not match the schedule. The balancer set the room offsets from the floor plan values. Where the full issued-for-construction (IFC) schedule set was not in hand, the only design reference was the TAB report’s own design column, which amounts to a report grading its own work. The chilled water differential pressure setpoint also appeared three ways across the report: 13 psi, 9.08 psi and a 40 psi plant value. The report needs to state the one setpoint the balancing was performed at, or the as-balanced condition cannot be reproduced during a re-balance or retrocommissioning.

Data that is too perfect
Readings that match design exactly call for a question about how they were measured, because real balance data scatters across the tolerance band. A common capture hood, the TSI Alnor EBT731, lists its flow accuracy as plus or minus 3 percent of reading plus or minus 7 CFM above 50 CFM. On a 100 CFM diffuser that is (0.03 x 100 CFM) + 7 CFM = 10 CFM, the entire plus or minus 10 percent tolerance band. A balancer also has no reason to keep adjusting once a reading falls inside tolerance.
On the campus project, 28 reheat coil circuit setters recorded actual flow matching required flow to the tenth of a GPM, with “NA” in every valve setting column. Automatic flow limiters would explain an exact match, in which case the report should identify them and show each pressure drop inside the valve’s control range. The recorded differentials ranged from about 18 to 68 psid against a 21 psi system setpoint, which also warrants a look at pump head and valve sizing. In the same report, an air handler chilled water coil flow of 115 GPM against a 109 GPM design was read from the BAS rather than measured at the valve.
On another project, roughly 250 energy recovery ventilator (ERV) grille readings were all exactly equal to design, with the face velocity column blank on every line. Rooftop unit sheets in the same report showed normal scatter, 345, 348 and 363 CFM against a 350 CFM design. So how likely is a column of perfect readings? For a 100 CFM grille recorded to the nearest whole CFM, a balancer stopping anywhere inside the plus or minus 10 percent band has 21 possible readings. Treating them as equally likely, the chance of all 250 grilles reading exactly design is (1/21)250 ≈ 10-331. Excel rounds that to 0, which is closer to the right answer than the report. Our finding stated that the ERV data appears to have been transcribed from design. We asked for the raw field sheets.
Know the device
We identify the device type before calling a low reading a deficiency, which is why our Rev 1 review on the laboratory building withdrew one of our own findings. The lab supply and exhaust valves read under design on the TAB sheets, and our first read was maldistribution calling for a floor re-balance. The valves were measured-airflow control valves, commanded from minimum to maximum by the room controller for sash position, occupancy and pressurization. The TAB agency does not field-balance a valve like that. A single reading at an unknown command state is not an acceptance basis for it. The controls startup report showed every valve good on airflow, actuator, sensor and communications. Acceptance comes from a functional test that commands each valve to minimum and maximum, runs sash tracking and face velocity, and confirms room pressurization.
The fume hood readings on the same building required the same distinction between balancing and controls. One hood came in about 14 percent below the 100 FPM face velocity requirement at the tested sash height. For a 6-foot hood at an 18-inch sash, the required exhaust is (9 sq.ft. x 100 FPM) = 900 CFM. At 14 percent low, the hood exhausts roughly 774 CFM, which is a containment concern. Because the hood valves track the sash, the fix is calibration of the sash sensor and face velocity setpoint by the controls contractor, not damper balancing.

Beyond balancing
A TAB report also exposes design and installation defects that no amount of balancing can correct. On the laboratory building, the numbers exposed several:
Corrosive cabinet exhaust connections choked down to a 1-inch reducer at the inlet read about 6 and 15 percent of design and pulled one exhaust branch down to 77 percent.
One air handler had no manual volume dampers installed, with the balancer’s note reading “unable to proportionally balance airflow.”
One exhaust valve topped out at 315 CFM where 425 CFM was required.
Hood tags had been switched in the field, with the schedule itself showing an 8-foot hood and a 10-foot hood swapped.
The controls startup report noted several chemistry lab exhaust valves “set up in analog mode, alarms turned off.” A disabled low-flow alarm on lab exhaust is a safety item that only appears when the controls startup report is read next to the TAB report.
These items go back to the contractor and the engineer, not the balancer.
Operating conditions
Three operating conditions on the campus project limited what the readings could prove: a temporary plant setpoint, the diversity method and the outside air mode. The chilled water side of one building was balanced before the central plant was fully online, using a temporary 18 psi differential pressure setpoint created by opening air handlers and fan coils in two other buildings. The balancer disclosed the workaround. Every chilled water flow in that report stays provisional until it is re-verified at the design setpoint.
On one air handler designed at 75 percent of installed terminal capacity, the balancer set the fan total by holding nine boxes at minimum and the rest at maximum, then added an 800 CFM offset on a return air box to hit the building pressure target. The method was sound and disclosed. The offset only holds if the controls contractor writes it into the permanent sequence. Building pressure also read 0.035 in. w.g. in the field against 0.025 in. w.g. at the BAS sensor that controls the return fan.
Minimum outside air on one air handler measured 2.3 percent short in minimum mode and 6.8 percent short in demand control ventilation mode, against a specification that allowed no negative deviation. That shortfall is a Title 24 ventilation compliance item. The outside air traverse also applied a 1.78 correction factor with no stated basis.
Coverage and the discrepancy log
Simple counts of equipment, test sheets and disclosed variances produced some of the most useful findings on the campus project:
An exhaust fan in the equipment register had no test sheet. Two heat exchangers on the water diagram at 10 GPM each appeared in no schedule or test sheet.
Two temperature test sheets carried another building’s name in the title block.
One VAV schedule had a minimum airflow higher than its maximum, a sequence no controls contractor can program.
Final damper and balancing valve positions were not marked or documented as the specification requires.
Our own equipment baseline needed the same scrutiny. The drawing extract showed 50 assets in one building where the real count was 43, because a second schedule table repeated seven VAV tags.
The balancer’s discrepancy log is a starting point rather than the review. Everything the campus balancer disclosed checked out. Several variances outside tolerance were not called out, including the outside air shortfall, the supply fan static, the 28 outlets and the exhaust fan horsepower. Both the firm and lead technician certifications also expired at the end of the year. Any re-balance carried past that date needs renewed credentials on the revised submittal.

Field verification
Field spot checks that repeat the conditions of the original balance confirm what a desk review cannot. We commonly verify a random 10 percent sample of terminals and major equipment, with each major system type represented. The spot check returns the system to the exact operating mode in which it was balanced, uses the same technicians and calibrated instruments where possible, and accepts deviations of roughly 10 percent or less for most air and water readings. We also observe at least one of each major equipment type and confirm that dampers and valves are marked or locked where required.
TAB should be substantially complete before intensive functional testing begins. Phased reports bring issues forward while the construction team is still mobilized.
How to review a TAB report
Start with the governing specification - Confirm which project sections apply and what they prohibit.
Read the field notes first - The balancer usually records the root cause.
Check every outlet against its own design value - A passing box total does not show that the room gets its air.
Reconcile schedule, TAB and BAS - Compare against the IFC schedules and get the engineer’s governing value in writing for every mismatch.
Check how the number was hit - Design CFM at 73 Hz is a flag and not a pass.
Ask how a perfect reading was measured - An exact match to design needs a stated method and the raw field sheets.
Identify the device first - A measured-airflow control valve is accepted by functional test, not by a balancing snapshot.
Read the controls startup report alongside - Disabled alarms and analog-mode valves appear there, not on the balancing sheet.
Record the operating conditions - Flag temporary setpoints, verify outside air in every control mode and carry offsets into functional testing.
Make an independent pass - Verify the discrepancy log, coverage and credentials, then spot-check a sample in its balanced mode.
Separate what balancing can fix - Missing dampers, undersized ducts and restrictions go to the contractor and the engineer.
The practical stakes
The TAB report is the last measured record of how air and water move through the building before the owner takes it over, which makes the review the last inexpensive chance to find a closed damper or a box programmed to the wrong maximum. Most of the TAB work on both projects was solid. The findings sat in the field notes, the device types, the fan speeds, the disabled alarms and the gap between the floor plan and the schedule, where a review that only lines up measured against design would not look.
On the campus project, the same SkySpark trend data we use for monitoring-based commissioning closed the loop during the warranty period. Comparing BAS setpoints against the schedule and TAB values for each box made clear which comfort calls were balance and design issues and which came from the controls.
References
NEBB Procedural Standard for Testing, Adjusting and Balancing - https://www.nebb.org
AABC National Standards for Total System Balance - https://www.aabc.com
ASHRAE Standard 110, Methods of Testing Performance of Laboratory Fume Hoods - https://www.ashrae.org
ASHRAE Guideline 0, The Commissioning Process - https://www.ashrae.org
TSI Alnor EBT731 Balometer specifications - https://www.instrumart.com/products/37678/tsi-alnor-ebt731-balometer
GMC Cx, Best Practice for Design and Submittal Reviews - https://www.gmccx.com/post/best-practice-for-design-and-submittal-reviews




Comments