Inputs
Formula used
This calculator compares a reference surface (spec/target) with a measured surface (field result) across multiple parameters. Each parameter is converted into a 0–1 similarity score and then combined with weights.
How to use this calculator
- Choose a similarity method and tolerance type.
- Enable the parameters relevant to your inspection scope.
- Enter reference targets and field measurements for each parameter.
- Set tolerances and weights to reflect project acceptance priorities.
- Click Calculate to view the overall similarity and per-parameter breakdown.
- Use the export buttons to download CSV or PDF records.
Example data table
The sample below shows one possible set of reference and measured values for a slab finish check. Your project tolerances and weights may differ.
| Parameter | Reference | Measured | Tolerance | Weight |
|---|---|---|---|---|
| Surface Roughness (Ra) | 8.0 µm | 9.2 µm | 20% | 2 |
| Flatness Deviation | 3.0 mm | 3.8 mm | 25% | 2 |
| Compressive Strength | 30 MPa | 28.5 MPa | 10% | 3 |
| Moisture Content | 4% | 5% | 30% | 1 |
Professional guide to surf similarity scoring
1) Why similarity matters on active sites
Surface acceptance is rarely a single-number decision. Finishes can meet strength yet fail bond, or look uniform yet exceed moisture limits. A similarity score helps teams compare many checks at once, document tradeoffs, and reduce rework risk when schedules are tight.
2) Common data sources and sampling
Typical inputs come from profilometers (Ra), straightedge or laser flatness readings, texture depth gauges, cylinder or core tests for strength, handheld moisture meters, tribometers for friction, absorption tests, and colorimeters for ΔE. For consistency, sample at the same grid spacing and record instrument model and calibration date.
3) Interpreting roughness and texture depth
Roughness (Ra) is often reported in micrometers, while texture depth is commonly reported in millimeters. For coated systems, small changes can affect adhesion. If your reference Ra is 8 µm and measured Ra is 9.2 µm, a 20% tolerance yields an absolute allowance of 1.6 µm under percent-baseline mode.
4) Flatness and levelness in practical terms
Flatness deviation is sensitive to the measurement length. When tolerances are strict, weight flatness higher to reflect downstream impacts such as flooring lippage and equipment alignment. Many projects track repeated readings to confirm finishing crews stay within a stable trend, not a single lucky point.
5) Strength and moisture thresholds
Structural and durability performance typically rely on minimum strength. Moisture limits often drive flooring and coating readiness; a small moisture overage can cause blistering later. Use higher weight for parameters that trigger stop-work decisions, and lower weight for aesthetic metrics when schedule pressure is high.
6) Friction coefficient and absorption for safety
Slip resistance is often evaluated using a friction coefficient. If your target is 0.60 and measured is 0.55, a 15% tolerance is tighter than it looks because the absolute allowance is only 0.09. Absorption is a durability proxy; higher absorption can indicate porosity and increased water ingress in harsh exposures.
7) Choosing the right method: linear, Gaussian, or cosine
Linear similarity is straightforward for acceptance checks. Gaussian similarity is useful when small deviations should not dominate the score, especially during early trial pours. Cosine similarity compares the overall pattern of values, helping when you care about profile “shape” more than strict per-parameter limits.
8) Using weights to reflect project priorities
Weights translate site priorities into the final score. A common approach is to weight strength and moisture at 3 each, flatness and roughness at 2 each, and aesthetics at 1. After each pour, compare scores across zones to identify crews, curing conditions, or equipment setups that consistently improve outcomes.
FAQs
1) What does a 90% similarity mean?
It means the weighted comparison is very close to targets within the selected tolerances. Review the per-parameter table to confirm that critical items like strength and moisture are not being masked by less important metrics.
2) Should I use percent or absolute tolerance?
Use percent tolerance when values scale with magnitude and you want proportional allowances. Use absolute tolerance when specifications state fixed limits in the same unit, such as ±0.5 mm flatness over a given length.
3) How do I set weights effectively?
Start by ranking parameters by risk: safety, durability, and compliance first. Assign larger weights to stop-work or warranty drivers. Keep weights simple, such as 1–3, so field teams can apply them consistently.
4) When is the Gaussian method better?
Gaussian is helpful for trial pours, benchmarking, and continuous improvement, because it reduces abrupt score drops for minor deviations. It still penalizes larger differences strongly, which keeps the score meaningful.
5) What does cosine similarity tell me?
Cosine similarity compares the overall direction of the reference and measured vectors. It is best when you want a holistic profile match across many parameters, not strict threshold acceptance for each parameter.
6) Can I export results for submittals?
Yes. Use the CSV export for spreadsheets and trend charts, and the PDF export for a quick field report. Exports include method, tolerance settings, overall score, and the parameter-by-parameter breakdown.
7) Why do I see the same similarity per row in cosine mode?
Cosine produces a single vector-based similarity. To avoid misleading row-level interpretations, the table displays the overall cosine score for each row. Use the reference and measured values to see which items drive the vector match.