Library / Dimensional analysis and similarity / DIM-01
Dimensional analysis and similarity
Number of dimensionless groups (Buckingham pi theorem)
Number of independent dimensionless groups needed to describe a problem with n physical variables (including the one you want to find), when their dimensions are built from r independent base dimensions.
- Ref
- DIM-01
- Rev
- v1
- Field
- Dimensional analysis and similarity
- Result
- Nπ
Before using any formula, check dimensional homogeneity: every term that is added or equated must have the same dimensions, written with mass M, length L and time T. For drag, FD = kρV²A gives [ρV²A] = (M L−3)(L² T−2)(L²) = M L T−2 = [FD], so the form is consistent and k must be dimensionless. The Buckingham π theorem then says that a physical relationship among n variables, whose dimensions are built from r independent base dimensions, can be rewritten as a relationship among Nπ = n − r independent dimensionless groups. Count every variable including the response, and take r as the rank of the dimension matrix, which is the number of independent dimensional combinations the variables represent. For drag on a body in a steady incompressible flow (force, density, speed, length and viscosity) n = 5 and r = 3, so there are two groups: a force coefficient and the Reynolds number.
Number of independent dimensionless groups needed to describe a problem with n physical variables (including the one you want to find), when their dimensions are built from r independent base dimensions. The Buckingham π theorem says the relationship can be rewritten using only these groups. Usually r is 3 (mass, length, time), or 4 when temperature enters.
Calculator
- Dimensionally independent repeating variables needed, r3
- Groups if one more variable is added3
- Groups if the same variables use one fewer base dimension3
- Share of the variables that become groups, Nπ/n40 %
What drives the result
Each variable is moved across its typical range while the others stay at the values above. The longest bar is the key driver.
- With 5 variables and 3 independent base dimensions, the problem reduces to 2 dimensionless groups: the original relationship among 5 quantities becomes one among 2 numbers, which is far easier to test and to plot.
- For drag on a body (drag force, density, speed, length and viscosity: n = 5, r = 3) there are two groups, a force coefficient FD/(ρV²ℓ²) and the Reynolds number ρVℓ/μ. A model test that matches the Reynolds number therefore gives the same force coefficient.
Chart data as tables
Ranked variables
| Variable | Typical range | Lowest result | Highest result | Swing () | Local step | Change in result |
|---|---|---|---|---|---|---|
| n Number of physical variables, including the response | 4 to 15 | 1 | 12 | 11 | +1 | +50 % |
| r Rank of the dimension matrix (independent base dimensions) | 1 to 4 | 1 | 4 | 3 | +1 | −50 % |
Sweep of Number of physical variables, including the response
| Number of physical variables, including the response | Number of independent π groups |
|---|---|
| 4 | 1 |
| 5.375 | 2.375 |
| 6.75 | 3.75 |
| 8.125 | 5.125 |
| 9.5 | 6.5 |
| 10.88 | 7.875 |
| 12.25 | 9.25 |
| 13.63 | 10.63 |
| 15 | 12 |
Sweep one variable
Variables and how each one contributes
Number of independent π groups
What the equation gives you. Everything else feeds it.
Number of physical variables, including the response
The number of physical variables in the problem, counting the response you want to predict (such as the force) as well as the inputs. Each extra variable adds one more group, so more variables mean more dimensionless numbers to control. Raising it by 1 (to 6) moves the result by +50 %, from 2 to 3.
Rank of the dimension matrix (independent base dimensions)
The rank of the dimension matrix: the number of independent base dimensions the variables use. It is 3 for mechanical problems in M, L and T, and 4 when temperature also appears. Each extra independent dimension removes one group, so more base dimensions give fewer groups. Raising it by 1 (to 4) moves the result by −50 %, from 2 to 1.
Worked example with the values above
- Count the variables, including the response: n = 5.
- Find the rank of the dimension matrix, the number of independent base dimensions (for example M, L, T): r = 3.
- Nπ = n − r = 5 − 3 = 2 independent dimensionless groups.
Assumptions and limits
- Every relevant variable is included and the relationship is dimensionally homogeneous.
- The rank r is that of the dimension matrix, which equals the number of base dimensions only when those dimensions are independent in the chosen variables.
Common mistakes
- Forgetting to count the response (such as the drag force) as one of the n variables.
- Assuming r = 3 every time: if no mass appears in the variables the rank is lower, and if temperature enters it can be 4.
- Treating the groups as unique: any product or power of groups is also a valid group, and the named numbers (such as the Reynolds number) are chosen for convenience.
Statistics and uncertainty
The count is exact for a given n and r, so the number itself carries no uncertainty. The risk lies in the inputs: leaving out a relevant variable gives too few groups and loses a needed ratio, while including an irrelevant one adds a group that does nothing. The theorem does not say which variables matter; that comes from the physics or from experiments.
Where it is used
- Planning scale-model, wind-tunnel and water-tunnel tests: deciding which numbers must be matched.
- Reducing the number of experiments or simulations needed to map a response.
Maths you need
- Dimensions: mass M, length L and time T, with every unit built from them
- Dimensional homogeneity: every term in a valid equation has the same dimensions