Parametric Costing

Estimate from drivers, not parts: cost per kilogram, per watt, per square meter — calibrated on past projects.

Time1 h
FormatSolo
StageToolkit

Parametric Costing: what it is and why it works

Parametric Costing estimates cost from a small number of physical or functional drivers, such as weight, power, area, capacity or part count, using a relation calibrated on past projects. The relation, often called a cost estimating relationship, is typically a power law of the form cost = a x driver^b, fitted by regression on the logarithms of historical data after normalizing for inflation, currency and scope. The estimator checks the scatter to judge the relation's reliability, applies it to the new product's drivers, and records the relation together with its validity range and the data it came from.

The method works because in early phases, when drawings are incomplete, a detailed should-cost model is too heavy and too speculative, while a calibrated driver relation captures the organization's real cost behavior. Parametric estimates are honest about their basis and fast to update as the design evolves. In process plants, the familiar capacity-scaling exponent, often around 0.6 for whole units, is a parametric relation of this kind. Parametric estimating is well established in AACE International practice and NASA cost-estimating handbooks. It depends on a clean Cost Database, can incorporate a Learning Curve for quantity effects, and produces estimates that should be graded as analogous (B) under Estimate Confidence Grading until replaced by detailed models.

What you need

Historical cost and driver data from comparable past projects
Normalization indices for inflation, currency and location
A candidate driver that plausibly explains cost
The new product's driver values and any known differences from the history

What you get

A fitted cost estimating relationship with its coefficients
A measure of scatter, such as R-squared or percentage error, and outlier notes
The estimate for the new product with a range
A database record: relation, data set, validity range, date

When to use it

When a full should-cost model is too heavy for the current phase.

How to do it, step by step

Pick the cost driver: weight, power, area, number of parts.
Collect historical cost-driver pairs from past projects.
Fit a simple relation and check the scatter.
Estimate the new product from its drivers with the fitted relation.
Record the relation and its validity range in the database.

Worked example: Estimating stainless process tanks from shell weight

Illustrative scenario — figures are realistic but not from a real company.

An engineering contractor needed early estimates for 11 stainless-steel process tanks in a food ingredients plant expansion. Only preliminary sizes were available, and the client wanted a budget within two weeks.

The estimator pulled 18 tanks fabricated for past projects, from 800 to 12,000 lb of shell weight, with their fabricated costs escalated to current dollars using a metals and labor index.
Plotting log cost against log weight showed a clear line. The regression gave cost = 210 x weight^0.78, with an R-squared of 0.92. Two tanks with jackets sat well above the line and were set aside to build a separate jacket adder.
The new tanks' shell weights were estimated from preliminary dimensions and standard thicknesses, then run through the relation, with the jacket adder applied to four of them.
The relation was recorded with its validity range, 800 to 12,000 lb, unjacketed 304 and 316L, atmospheric to low pressure.

Result. The parametric budget came to $1.36 million with a stated range of about plus or minus 20%. Vendor quotes received three months later totaled $1.29 million, within the range. The relation was added to the cost database and refitted with the new data points.

Open it in the FieldKit Design to Cost app → Get the Design to Cost template pack — join the waitlist

Common pitfalls and how to avoid them

Using data from different years and currencies without normalization.Escalate all historical costs to a common date and currency with appropriate indices before fitting.
Applying the relation outside the range of the data.Record the validity range and flag any estimate that extrapolates beyond it.
Choosing a driver because it is available rather than because it drives cost.Check that the driver has an engineering link to cost and test alternatives on the scatter.
Mixing designs with different features in one data set.Separate features that shift cost, such as jackets or pressure ratings, into adders or distinct relations.

Frequently asked questions

What is a cost estimating relationship?

A cost estimating relationship, or CER, is a mathematical equation that predicts cost from one or more technical or programmatic variables, such as weight, power or capacity. It is derived from historical data, usually by regression, and documented with its data set, statistics and range of validity. CERs are the building blocks of parametric estimating.

What is the six-tenths rule in cost estimating?

It is a rule of thumb for scaling the cost of process equipment or plants with capacity: cost ratio equals capacity ratio raised to an exponent, often around 0.6. Actual exponents vary by equipment type and range, so the rule is suitable for screening estimates. Where you have your own historical data, a fitted exponent is better.

How accurate is parametric estimating?

Accuracy depends on the quality and relevance of the historical data, the strength of the driver-cost link and whether the new item lies within the data range. Measure it directly: report the scatter of the fitted relation and check estimates against later actuals. Treat parametric results as early-phase estimates and replace them with detailed models as design matures.

Origin

Parametric cost estimating — AACE International; NASA cost-estimating handbook lineage.

Related methods

More in “Toolkit”

Cross-cutting methods: curves, complexity, quality and estimating tools.