Warehouse Slab Flatness and Levelness in 2026: What Your Racking Actually Requires, What Superflat Costs, and Who Pays When It Misses
Warehouse · Concrete Slabs · Updated August 2026
Warehouse Slab Flatness and Levelness in 2026: What Your Racking Actually Requires, What Superflat Costs, and Who Pays When It Misses
FF/FL and F-min are two different measurement systems that do not convert to each other, and specifying the wrong one is how a passing floor turns into a $1 million dispute after the racking arrives. Here are the real numbers by use case, the cost by tier, and the twelve line spec that keeps you out of it.
The Short Answer
FF35 / FL25 overall with FF21 / FL15 minimum local is the defensible 2026 default for a random traffic warehouse, and it matches ACI 301's industrial floor default. If any aisle runs VNA or turret trucks above roughly 35 feet, F-numbers are the wrong system entirely and you need an F-min spec at 60 to 100 longitudinal and 70 to 125 transverse, keyed to rack height. Superflat labor runs 50 to 100 percent above conventional, but the number that actually controls your budget is area, not rate. Superflat only needs to exist in the wheel paths. Specifying it building wide instead of aisle by aisle is the single most expensive mistake on this list.
Key Takeaways
- FF/FL and F-min do not correlate. Not directly, not approximately. A floor that passes FF50/FL35 with room to spare can still fail F-min 75 in an aisle.
- The 72 hour test window is an ACI 117 requirement, not an ASTM one. Specs that cite only ASTM E1155 have no timing requirement and are legally weak.
- Documented curl losses: one industrial warehouse went from FF 68.7 to 41.4 and FL 73.4 to 36.8 over twelve months. The higher your initial F number, the more of it curl destroys.
- Grinding fixes FF, not FL. FF is 24 inch curvature. FL is a 10 foot elevation trend. A slab that fails FL badly usually cannot be ground into compliance.
- Cross aisle level error is magnified at the mast top in direct proportion to mast height. At a 44 inch wheel track lifting to 40 feet, that is roughly an eleven to one multiplier.
- Freezer slabs contract three to four times more than dry slabs and sit on a slip sheet with almost no subgrade friction. A joint sawn at 1/8 inch routinely opens to 3/8 inch or more at operating temperature.
- Omitting the minimum local value is the most consistent specification error in the industry. Default is 3/5 of the specified overall value, and never below FF13 / FL10.
What This Covers
- The four systems, and which one applies to your building
- The numbers by use case
- Why a turret truck sways when the floor passed inspection
- What flatness costs by tier in 2026
- What remediation costs, and what grinding cannot fix
- The rework dispute, modeled
- Freezer slabs: why they move differently
- Who owns the risk in design build versus design bid build
- The twelve line spec
- Frequently asked questions
Four Systems, And Which One Applies
Most cost overruns on warehouse slabs trace back to a specification that used the wrong measurement system for the equipment that eventually moved in. Here is the landscape.
| System | Standard | How it measures | Use it for |
|---|---|---|---|
| FF / FL | ASTM E1155/E1155M-23, tolerances per ACI SPEC-117-10(15) | Random sample lines, readings at 12 in. max. FF from curvature over 24 in. FL from elevation difference over 10 ft. Statistical, 90 percent confidence | Random traffic. Wide aisle warehouse, offices, retail, cross dock |
| F-min | Face Companies proprietary, table in ACI PRC-360-10 | Profileograph running the actual wheel paths, at point separations set by that truck's wheelbase and load wheel track. Reports the worst condition, not an average | Defined traffic. VNA, turret truck, wire or rail guided, floor running stacker cranes |
| Waviness / wheel path | ASTM E1486/E1486M-14(2022) | Waviness index, transverse and longitudinal wheel path elevation difference, RMS slope | The ASTM route to defined traffic where a proprietary system is not acceptable |
| TR34 FM / DM classes | Concrete Society TR34, 4th edition | 95th percentile limits in millimeters over 3 m and 300 mm | European specs, international 3PL tenants, anything written in millimeters |
The Conversion That Does Not Exist
There is no correlation, direct or approximate, between FF/FL and F-min. They measure different geometry on different sample paths with different statistics. F-min cannot even be averaged, because it reports the worst acceptable condition in the wheel path. If a vendor hands you a conversion table between the two, it has no standing in a dispute and it will not help you when the turret truck sways. The same applies to converting TR34 FM classes into F numbers. If the lease says FM2, you have to measure FM2.
The ACI 117 Tolerance Table
| Classification | Specified overall FF | Specified overall FL | Min local FF | Min local FL |
|---|---|---|---|---|
| Conventional | 20 | 15 | 12 | 9 |
| Moderately flat | 25 | 20 | 15 | 12 |
| Flat | 35 | 25 | 21 | 15 |
| Very flat | 45 | 35 | 27 | 21 |
| Super flat | 60 | 40 | 36 | 24 |
Two rules travel with that table. The theoretical flat plane the floor is measured against must itself be within 3/4 inch of design elevation. And FL is not specifiable on inclined or cambered surfaces, or on unshored elevated slabs. That last one matters more than people expect, because it means FL is unenforceable on your dock apron, on any slab sloped to drains, and in most food processing wash down areas.
The Numbers By Use Case
| Use case | System | Specified overall | Minimum local |
|---|---|---|---|
| Loading dock and cross dock apron | FF only plus absolute elevation at the pit rim | FF20 / FL15 | FF12 / FL9 |
| Standard random traffic warehouse, 12 ft aisles and up | FF/FL | FF35 / FL25 is the 2026 market default | FF21 / FL15 |
| Narrow aisle 8 to 10 ft, reach truck to ~30 ft | FF/FL | FF35 / FL25 minimum, FF45 / FL35 better in aisles | FF27 / FL21 at the higher tier |
| VNA, turret truck, wire or rail guided, above ~35 ft | F-min only | F-min 60 to 100 longitudinal, 70 to 125 transverse by rack height | Not applicable, F-min is worst case |
| Superflat, defined traffic top tier | F-min | F-min 100 | Not applicable |
| AS/RS, floor running stacker cranes | F-min in the runway, FF/FL elsewhere | Per ACI 360R by rack height. Above 90 ft the crane OEM tolerance governs | Not applicable |
| AMR and AGV | FF/FL plus a joint condition spec | FF35 to FF50 / FL30 to FL40 typical | 3/5 of overall |
| Freezer below 0 F | FF/FL, tested twice | FF35 / FL25 typical | FF21 / FL15 |
| Food processing, production with slope | FF/FL | FF30 / FL25 | FF18 / FL15 |
| Food processing, washdown | FF only plus slope | FF25 / FL20 | FF15 / FL12 |
| Precision manufacturing and metrology | FF/FL | FF45 to FF60 / FL35 to FL40 | 3/5 of overall |
| Broom finish, anywhere | FF only | FF15 to FF20 realistically, no matter what the spec says |
Two things on that table deserve emphasis. First, the broom finish line is not a joke. The American Society of Concrete Contractors documents that broom finished surfaces measure FF15 to FF20 regardless of what the specification demands, and that specifiers routinely write FF20 to FF35 on surfaces a broom physically cannot produce. Second, the AMR and AGV row is the fastest moving area of this whole subject, and there is no settled industry standard yet. Write the joint condition, not just the F number.
The F-min Table By Rack Height
| Rack height | Longitudinal F-min | Transverse F-min |
|---|---|---|
| 0 to 25 ft | 50 | 60 |
| 26 to 30 ft | 55 | 65 |
| 31 to 35 ft | 60 | 70 |
| 36 to 40 ft | 65 | 75 |
| 41 to 45 ft | 70 | 80 |
| 46 to 50 ft | 75 | 85 |
| 51 to 65 ft | 90 | 100 |
| 66 to 90 ft | 100 | 125 |
Why A Turret Truck Sways When The Floor Passed Inspection
The physics is geometric, not empirical, and it explains the entire transverse column in that table.
TR34 states it plainly: variation in floor level across an aisle between the wheel tracks of material handling equipment is magnified at the top of the mast in direct proportion to its height, and variations in level also induce dynamic movement that magnifies the static lean.
The multiplier is mast height divided by wheel track width.
Worked Example
A typical VNA turret truck has a wheel track around 44 inches and lifts to 40 feet. The multiplier is 480 divided by 44, or roughly eleven to one. A 1/8 inch transverse elevation difference between the two wheel tracks produces roughly 1.4 inches of lateral fork displacement at 40 feet, before any dynamic sway is added. At 60 feet the same 1/8 inch produces about 2 inches. This is Terrapin's derived calculation from the TR34 relationship, not a published figure.
That is why cross aisle error leans the mast and why transverse F-min is always the higher number. It is also why an owner who specified FF/FL for a VNA building was never measuring the thing that causes the problem.
What Out Of Spec Actually Costs In Operations
- Mast sway and fork misalignment at high lift
- Loss of wire guidance signal or outright tracking failure
- Forced deceleration and reduced travel speed, which directly reduces throughput
- Accelerated wear on wheels, axles, and control electronics. Documented forklift steer axle repair attributable to floor impact runs $1,200 to $4,500 per unit per year
- Rack impact damage from steering misalignment
- 15 to 20 percent reduction in picks per hour on rough floors
That last figure is the one to put in front of a 3PL tenant. A 15 to 20 percent throughput haircut is not a maintenance issue. It is a rent renewal issue.
The Amplification Problem
One Eighth Of An Inch, Amplified
Lateral fork displacement at the top of the mast from a 1/8 inch transverse difference between wheel tracks. Multiplier is mast height divided by wheel track width, taken at 44 inches.
Terrapin Construction Group derived calculation from the TR34 relationship quoted in Concrete NZ Technical Specification 01:2021. Static displacement only, before dynamic mast sway.
What Flatness Costs By Tier
Baseline first. A 6 inch warehouse slab for light to medium use runs roughly $7.50 to $11.00 per square foot in 2026, and an 8 inch or thicker heavy industrial slab runs roughly $11.00 to $16.50, excluding site prep, subgrade, vapor barrier, and surface treatments.
Two hard anchors exist in the published literature for what flatness adds. A Fricks superflat case study published through ACI documented that labor cost for superflat work ran 50 to 100 percent greater than conventional slab work, because crew productivity for forming and finishing was cut by one third to one half. Separately, published 2026 guidance puts the total floor cost premium for higher FF/FL at 10 to 25 percent.
| Tier | Spec | Incremental over conventional | Applied to |
|---|---|---|---|
| 0 | FF25 / FL20, ACI 302 Class 4 or 5 | Baseline | Whole floor |
| 1 | FF35 / FL25, ACI 301 industrial default | +$0.15 to $0.40/SF | Whole floor |
| 2 | FF45 / FL35, very flat | +$0.50 to $1.25/SF | Whole floor |
| 3 | FF60 / FL40, superflat random traffic | +$1.25 to $2.50/SF | Whole floor |
| 4 | Defined traffic, F-min 50 to 75 | +$1.50 to $3.00/SF | Aisle area only |
| 5 | Defined traffic, F-min 100 and up, strip pour | +$3.00 to $6.00/SF | Aisle area only |
The Cost Point That Matters Most
Superflat is an aisle area cost, not a building area cost. Defined traffic F-min applies only to the wheel paths. A 500,000 square foot distribution center with 120,000 square feet of VNA aisle should buy Tier 5 on the 120,000 and Tier 1 on the balance, not Tier 5 on 500,000. Owners who let the spec run building wide pay a seven figure premium for flatness in areas no wheel will ever touch.
Laser Screed Versus Manual, And Which Number Each Buys
Somero's framing is the technically correct one and it should govern how you write the scope: FF is a function of finishing. FL is a function of forming and strike off. You cannot trowel your way to FL. Laser screed is what buys FL.
| Method | Throughput | Crew | 10,000 SF pour |
|---|---|---|---|
| Laser screed | 2,500 to 3,000 SF/hr practical | 1 operator plus 4 to 5 support | 3 to 4 hours |
| Skilled manual crew | 500 to 800 SF/hr | 8 to 12 finishers | Multiple days |
Two low cost levers worth writing into the scope. Replacing bullfloats with highway straightedges yields a 25 to 50 percent increase in FF at minimal cost. And reducing form or screed spacing below 25 feet is what physically enables FF above 40. If your spec asks for FF45 and your scope allows 40 foot screed rail spacing, the spec and the means are in conflict before anyone shows up.
What Remediation Costs, And What Grinding Cannot Fix
| Method | 2026 cost | What it actually corrects |
|---|---|---|
| Light diamond grinding, surface profiling | $1.50 to $3.00/SF | Surface profile only |
| Slab leveling, high spot and curled joint grinding | $2.50 to $4.50/SF | FF. This is the F number correction line |
| Heavy grinding, coating and mastic removal | $3.50 to $6.00/SF | Surface preparation |
| Self leveling underlayment, installed | $3.40 to $9.20/SF, large open areas $2.50 to $5.50 | Localized elevation. Not rated for hard wheel forklift traffic |
| Slab removal and disposal | $2.00 to $5.00/SF | |
| New replacement slab | $8.00 to $15.00/SF | |
| Full removal and replacement, all in | $10.00 to $20.00/SF | Everything |
| Urethane cement topping, freezer, wet, FDA | $8.00 to $16.00/SF | Wear surface plus modest profile correction |
Grinding Fixes FF. Grinding Does Not Fix FL.
FF is 24 inch curvature and a grinder can take that out. FL is a 10 foot elevation trend. Correcting a long wavelength FL failure by grinding means removing an inch or more of slab across a large area, which removes cover, exposes aggregate and sometimes reinforcement, and destroys the wear surface. A slab that fails FL badly usually cannot be ground into compliance. That is the failure mode that ends in a bonded topping or in removal and replacement.
Second trap: residential grade self leveling underlayment is not rated for hard wheel forklift traffic. For a warehouse the correct product is a high strength bonded industrial topping, minimum 3/4 inch, per ACI 302 Class 7 logic. Pricing an SLU fix into a warehouse rework is a common and expensive estimating error.
The Rework Dispute, Modeled
Here is why the 72 hour test is not optional. Scenario: 100,000 square feet of VNA aisle fails F-min, the 72 hour survey was skipped, and the failure surfaces at rack install.
| Line item | Exposure |
|---|---|
| Grinding remediation at $3.50/SF over 100,000 SF | $350,000 |
| Re-survey, F-min profileograph, two rounds | $10,000 to $16,000 |
| Racking installer demobilization, remobilization, standby | $25,000 to $75,000 |
| Rack storage and detention on delivered but uninstallable steel | $15,000 to $50,000 |
| Rent commencement delay, 1 to 2 months at $0.60 to $0.90/SF/mo | $60,000 to $180,000 |
| Schedule impact | 3 to 6 weeks of continuous grinding |
| Realistic total exposure | $500,000 to $1.2M plus delay damages |
The mechanical reason it is this expensive: grinding cannot be performed under installed racking. If anchors are drilled and uprights are set, the rack has to come out, the anchor holes get ground through or abandoned, and the layout has to be re-set and re-anchored. There is no partial fix.
That is the whole argument for testing every pour at 72 hours instead of testing once at the end. A per pour survey turns a potential seven figure end of job dispute into a $15,000 conversation after pour three.
Freezer Slabs Move Differently
The specified number is often the same as a dry slab, typically FF35 / FL25. The behavior is not remotely the same, and a spec that does not account for that will end in a dispute.
Three Mechanisms That Compound
01
Near zero subgrade friction
The slab is cast on a slip sheet over rigid insulation, not on granular base. Panels are free to move. Joints open further and panels curl more freely than a comparable dry slab.
02
Thermal contraction is three to four times greater
The DoD UFC states it directly. A joint sawn at 1/8 inch routinely opens to 3/8 inch or more at operating temperature.
03
The slab cannot dry downward
Top down drying over an impermeable vapor barrier and insulation produces the classic curl gradient, and refrigeration accelerates surface moisture loss. Panel edges lift, the panel cantilevers off support, and under traffic the panel rocks, leaving joint edges elevated at exactly the moment a wheel hits them.
What Curl Actually Destroys
| Documented case | Initial | After | Loss |
|---|---|---|---|
| Pennsylvania industrial warehouse, 12 months | FF 68.7 / FL 73.4 | FF 41.4 / FL 36.8 | FF down 40%, FL down 50% |
| University gymnasium, 7 months | FF down 10 to 20% | ||
| Typical corner curl magnitude | About 1/4 inch at slab corners, less at joints |
The practical consequence for spec writing is unavoidable. A freezer superflat spec has to be tested at 72 hours for contractor acceptance, and re-surveyed at freeze down as a separate, separately priced scope with its own acceptance criteria and its own remedy. Anyone who writes one number and tests it once will end up in a dispute, and both parties will be partly right.
The Freezer Assembly
Bottom to top, per the DoD UFC: base courses, sub slab carrying heat return conduits, vapor retarder placed directly below the insulation, rigid insulation 4 to 8 inches, slip sheet, structural slab on ground.
| Item | Requirement | Why it matters |
|---|---|---|
| Vapor retarder | Less than 0.1 perms after conditioning | ACI 302 and ACI 360 now recommend one under all climate controlled cooled environments |
| Vapor barrier | 0.01 perms or less after conditioning | |
| Rigid insulation is not a vapor barrier | 1 inch of standard EPS is about 2 to 5.8 perms | Two orders of magnitude off. The barrier is a separate layer and goes below the insulation |
| EPS compressive strength | 30 psi commonly recommended for freezer service | |
| Under slab heat load | 1.5 to 2.5 BTU/SF/hr | Sized on soil to freezer temperature differential |
| Heat method | Glycol tubing is most economical. Electric heat tape is the retrofit method | Gravity and forced air duct systems have a history of failure and are not recommended |
| Subgrade modulus | Do not use the insulation supplier's k value | Supplier values run higher than plate bearing test values and produce an under thick slab. Use 75 to 100 pci if actual soil k is unknown |
| Perimeter insulation | 24 to 36 in. in from exterior foundation and walls in cold climates | Sometimes turned down vertically along the grade beam |
Joint Filler, And The Hold Points That Belong In Your Spec
| Property | Semi rigid epoxy | Polyurea |
|---|---|---|
| Initial set | 4 to 8 hours | 1 hour or less |
| Temperature range | 35 F and above only | Installs in sub zero conditions |
| Equipment | Manual guns or single pump | Requires dual component pump |
| Moisture sensitivity | Generally resistant | Can bubble or foam if the joint is damp |
| Sawable for repair | Yes | Not effectively |
Write these hold points into the specification verbatim:
- Coolers: hold the room at planned operating temperature for at least 48 hours before joint filling begins.
- Freezer rooms below 0 F: maintain operating temperature for 14 days before starting to fill joints.
- Delay joint filling as long as practically possible regardless.
- Specify joint fillers specifically developed for cold temperature applications.
- Steel armored joints 3/8 inch or wider must be filled full depth, to give wheels a smooth transition and protect tire tread.
- Include a contract provision requiring the contractor to return between 6 months and 1 year to repair joint filler separation. Separation is not by itself a defect. The accepted threshold is that voids should not exceed credit card width and the material must remain firmly seated.
Two Interfaces Nobody Specifies
The IMP wall base line. Insulated metal panels are rigid and dimensionally fixed. Slab levelness error at the base track does not get absorbed. It propagates up the wall as joint gap variation and shows at the roof line. There is no published slab flatness tolerance for the IMP base line, and we are not going to invent one. The practice recommendation is to specify a separate, tighter absolute elevation tolerance on the 4 to 6 foot strip at the wall line, independent of the FF/FL spec for the field of the slab, and to require a survey of the panel starting line before panel delivery. In freezers the base detail also carries the thermal break and the turn up of the under slab insulation and vapor barrier, so elevation error there breaks the thermal envelope at exactly the point where a short causes frost. We cover the base conditions in detail in the IMP installation guide.
Dock pit rims. Pit style levelers are set to finished floor elevation and the pit curb angle has to match it. Slab FL error at the dock face shows up as a step at the pit rim, which is a trip and wheel impact point, and as lip to trailer bed misalignment. The apron and the interior area in front of the doors are usually sloped, which means FL is not enforceable there. Specify FF plus an absolute elevation tolerance at the pit rim, referenced to the same datum as the leveler shop drawings. The dock side of the building is covered in the distribution center construction guide.
Who Owns The Risk
Design bid build. The engineer of record owns the specification. The GC and its concrete subcontractor own means, methods, and the resulting F numbers. If the spec is achievable and the floor misses, the concrete sub pays. If the spec is internally contradictory, the contractor has a real defense and the owner usually absorbs the cost.
The recurring contradictions, all documented by ASCC, are worth memorizing because they show up on most projects:
- Division 03 F numbers alongside a Division 09 "1/8 inch in 10 feet" straightedge requirement, with no reconciliation between them
- F numbers specified across construction joints, which ASTM E1155 explicitly excludes
- F numbers specified on broom finished surfaces at levels a broom cannot produce
- F numbers specified in congested areas where the finishing equipment that produces those numbers physically does not fit
Design build. The design builder owns both halves. There is no engineer to point at and no defective spec defense. The protective move is procedural rather than legal: get the F number spec, the system selection, the tier boundaries by area, and the testing protocol into the owner's program documents before the GMP is set. If the tenant is not identified at GMP, carry the flatness tier as a stated allowance keyed to a named material handling assumption, and put that assumption in the Owner Responsibility Matrix so a later tenant driven upgrade is a change order rather than a swallow. That is how we handle it on design build projects.
ASCC's recommendation applies in both delivery models and is worth adopting as house practice: carry a bid allowance, set by the design professional and based on the actual floor covering and equipment requirements, for the grinding and patching needed to close the gap between Division 03 tolerances and downstream tolerances, with unused allowance returned to the owner. It lets flooring and racking bids be compared on equal footing and it puts the gap on the table before anyone pours.
The Twelve Line Spec
None of these cost anything at design stage. All of them are expensive after the pour.
- Name the standard and the edition. ACI SPEC-117-10(15), ACI 301-20, ASTM E1155/E1155M-23, ACI PRC-302.1-15, ACI PRC-360-10, and ASTM E1486/E1486M-14(2022) where applicable.
- Specify both the specified overall value and the minimum local value. Default local is 3/5 of overall. Never below FF13 / FL10.
- Say which system applies to which area, on a keyed floor plan. FF/FL for random traffic, F-min for defined traffic aisles. Do not specify FF/FL and expect VNA performance out of it.
- State the theoretical flat plane rule. The floor must be within 3/4 inch of design elevation.
- State the testing window in the spec, not just by reference. Within 24 hours after placement where practical, and in no case later than 72 hours after completion of finishing operations.
- Name the corrective method in advance. Grinding, planing, surface repair, overlay, or removal and replacement. Say which, and say who chooses.
- Exclude the 2 foot band at construction joints, penetrations, and slab edges from the F number calculation, per ASTM E1155.
- Lower the requirement where the method cannot deliver it. Congested areas, ramps, broom finishes, and heavily reinforced slabs with thin cover. On heavily reinforced floors either lower the F number or increase cover to 1.5 inches.
- Require a preconstruction conference at least two weeks before the first slab pour, with the owner, design professional, GC, concrete contractor and finisher foreman, concrete producer, testing agency, and pumping contractor. Put the F number specification, the repair guideline, the window of finishability, curl mitigation, corrective actions, and dispute resolution on the written agenda.
- Require a mock up or first pour test section, surveyed within 72 hours, before the balance of the slab is released.
- Write the freezer hold points in. Cooler 48 hours at operating temperature before joint fill. Freezer below 0 F, 14 days. Cold service filler required. Return visit between 6 months and 1 year.
- Separate the curl question explicitly. State that F numbers measured after the 72 hour window are not the basis of acceptance, and that any later survey is a separate scope with its own criteria and its own price.
Get The Floor Spec Right Before The GMP
Terrapin Construction Group builds warehouse, cold storage, and distribution facilities nationwide, licensed in all 50 states. We set the flatness tier by area against the actual material handling assumption during preconstruction, so the superflat premium lands in the aisles and nowhere else.
Schedule A 30 Minute Call Get An Instant EstimateFrequently Asked Questions
What FF and FL do I need for a warehouse?
If it is random traffic with reach trucks and aisles 10 feet or wider, FF35 / FL25 overall with FF21 / FL15 minimum local is the defensible 2026 default, and that is also ACI 301's industrial floor default. FF25 / FL20 is the floor that will function but will not feel flat. If any part of the building runs VNA or turret trucks above about 35 feet, F numbers are the wrong system for those aisles entirely and you need an F-min specification.
How much extra does a superflat floor cost per square foot?
Superflat labor runs 50 to 100 percent above conventional slab labor because crew productivity is cut by a third to a half. In 2026 dollars that lands roughly at $1.25 to $2.50 per square foot over a conventional floor for random traffic superflat, and $3.00 to $6.00 per square foot for defined traffic F-min 100 strip pours. The number that actually controls your budget is not the rate, it is the area. Superflat only needs to exist in the wheel paths. Specify it aisle by aisle rather than building wide and the premium drops by 70 to 80 percent.
Can you just grind the floor flat if it fails?
Sometimes. Grinding corrects FF, which is short wavelength bumpiness over 24 inches, at roughly $2.50 to $4.50 per square foot for high spot and curled joint work. Grinding usually cannot correct an FL failure, which is a 10 foot elevation trend, because fixing it means removing an inch or more of slab across a large area, taking out cover and exposing aggregate. FL failures tend to end in a bonded topping or in removal and replacement at $10 to $20 per square foot.
When does the floor have to be tested?
Within 72 hours after finishing operations are complete, and preferably within 24. That is an ACI 117 requirement, not an ASTM one. It exists to separate what the contractor built from what curl and drying shrinkage do afterward. Documented cases show FF and FL dropping 40 to 50 percent over 12 months from curl alone, so a test taken at rack install measures a different floor than the one that was poured.
My 3PL tenant sent a spec written in millimeters. What is FM2 and how does it convert?
FM2 is a Concrete Society TR34 Free Movement class: 6.5 mm levelness over 3 meters and 2.0 mm flatness over 300 mm at the 95th percentile. It is the European default for reach truck distribution centers operating between 8 and 13 meters. It does not convert to FF/FL. It is a different measurement geometry with a different statistical basis, and any conversion table you are shown is an approximation with no standing in a dispute. If the lease says FM2, you have to measure FM2.
What is F-min and why can I not just use FF and FL for my VNA aisles?
F-min is the defined traffic system. It is measured with a profileograph running the actual wheel paths of the actual truck, at point separations set by that truck's wheelbase and load wheel track, and it reports the worst condition in the path rather than a statistical average. FF/FL samples the floor randomly and averages. There is no correlation between the two systems, direct or approximate. A floor that passes FF50 / FL35 with room to spare can still fail F-min 75 in an aisle, and the reverse is also true.
Why does my turret truck sway so much at height when the floor passed inspection?
Because cross aisle level difference between the two wheel tracks is magnified at the top of the mast in direct proportion to mast height. With a 44 inch wheel track lifting to 40 feet the multiplier is about eleven to one, so an eighth of an inch of difference between wheel tracks becomes roughly an inch and a half of lateral fork displacement at the top, before dynamic sway adds to it. If the floor was inspected against FF/FL rather than F-min, it was never measured for the thing that causes the sway.
Do freezer slabs need a different flatness spec than dry slabs?
The specified number is often the same, FF35 / FL25 is typical, but the behavior is completely different and the spec has to account for it. A freezer slab sits on a slip sheet over rigid insulation, so it has almost no subgrade friction, its thermal contraction is three to four times greater than a dry slab, and it cannot dry downward through the vapor barrier. Joints sawn at an eighth of an inch routinely open to three eighths or more, and curl is worse. Write a 72 hour acceptance test plus a separate, separately priced re-survey at freeze down, and do not fill joints until the room has held operating temperature for 48 hours in a cooler or 14 days in a freezer below zero.
Who pays when the slab is out of tolerance and my racking is already on site?
In design bid build, the concrete subcontractor pays if the spec was achievable and internally consistent, and the owner usually absorbs it if the spec contradicted itself. In design build the design builder owns both halves and there is nobody to point at. Either way the number is large, because grinding cannot be done under installed racking. The rack has to come out, anchors get abandoned, and the layout has to be re-set. On 100,000 square feet of failed VNA aisle, a realistic total including grinding, re-survey, installer remobilization, and one to two months of delayed rent commencement is $500,000 to $1.2 million.
What is the one thing I should add to my spec to avoid all this?
Specify both an overall and a minimum local F number, and require an ASTM E1155 survey within 72 hours on every single pour, paid by the concrete subcontractor and reported within 48 hours, with the corrective method named in advance. Omitting the minimum local value is the most common specification error in the industry, and a per pour 72 hour survey turns a potential million dollar end of job dispute into a $15,000 conversation after pour three.
Sources And Further Reading
- ACI SPEC-117-10(15), Specification for Tolerances for Concrete Construction and Materials
- ACI PRC-302.1-15, Guide to Concrete Floor and Slab Construction
- ACI PRC-360-10, Guide to Design of Slabs-on-Ground
- ACI 301-20, Specifications for Concrete Construction
- ACI excerpted tolerance resource WTRF6, free PDF with the SOFF and SOFL tables
- ACI 360R-10 free preview, confirming the refrigerated facilities and curling chapters
- ASTM E1155/E1155M-23, Determining FF Floor Flatness and FL Floor Levelness Numbers
- ASTM E1486/E1486M-14(2022), Waviness, Wheel Path and Levelness Criteria
- DoD UFC, Design of Concrete Floor Slabs-on-Ground for DoD Facilities, 1 March 2025
- WBDG Unified Facilities Criteria index
- US Access Board, Dimensional Tolerances in Construction and for Surface Accessibility
- NIH Office of Research Facilities, Concrete Slab Profile Quality technical bulletin
- ASCC Position Statement 4, Separation of Semirigid Concrete Floor Joint Fillers
- ASCC Position Statement 25, Floor Finishing Specifications
- ASCC Position Statement 6, Division 3 versus Division 9 Floor Flatness Tolerances
- ASCC, The Limitations of F-Number Specifications in Concrete Construction
- ASCC, Concrete Curling and the Effects on Floor Flatness and Levelness
- Allen Face, Concrete Slabs and the Use of the F-Number System
- Face Consultants, the ACI F-min Number System
- Face Consultants, TR34 Defined Movement classifications, 4th edition
- Concrete Society TR34, 4th edition
- Concrete NZ Technical Specification 01:2021, Surface Regularity Requirements
- Somero F-Numbers Handbook
- Metzger McGuire, Focus on Industrial Floor Joints
- Fricks, Constructing a Superflat Concrete Floor Slab
- Fricks, Why Refrigerated Warehouse Floors Require Vapor Barriers
- MHI Solutions Guide, Automated Storage and Retrieval Systems
Terrapin Construction Group is a nationwide design build commercial general contractor headquartered in Denver, Colorado, licensed in all 50 states, and a Procore Certified Contractor. We build warehouse, cold storage, food processing, and distribution facilities across the country, and we install commercial flooring systems including polyaspartic and urethane cement.
Cost figures are budgetary and current as of August 2026. Tier pricing in this article is modeled from published labor premium data and should be validated against live subcontractor pricing before it is committed to a guaranteed maximum price. Standard editions change. Confirm the current edition of every standard cited here before writing it into a specification. Nothing in this article is a substitute for a structural engineer's stamped slab design.
