About Research

Strength prediction

Given a mix design (cement, slag, fly ash, water, superplasticizer, aggregates, and curing age), the backend model predicts compressive strength in MPa. The model identifier returned with each prediction is shown alongside the result so predictions stay traceable to the model version that produced them.

CO₂ and cost estimation

CO₂ output and material cost are estimated per m³ from the same mix design, using a per-material breakdown so it is possible to see which components of a mix contribute most to emissions or cost.

Mix optimization (v0.1)

The optimizer is a weighted-sum SLSQP solver: it combines strength, cost, and CO₂ objectives into a single scalar objective and returns one best trade-off mix, subject to whatever constraints are set. This means v0.1 returns exactly one result, not a set of alternatives to choose between.

A future version is planned to add a Pareto-front optimizer (NSGA-II) behind a separate endpoint, returning multiple non-dominated candidates instead of one. That endpoint does not exist yet — this page will be updated when it ships, and it will be a new tool alongside the current one rather than a replacement for it.

Model Comparison

The Model Comparison page is reserved for benchmarking prediction models against each other (RMSE, MAE, R²). It stays empty until a comparison endpoint exists on the backend, rather than showing placeholder figures.