SoftwarePlus.ai

All programs › BMPFast

Constant Validation and Verification of BMPFast

How we make sure BMPFast follows the permitting rules and calculates them correctly — not once, but with every change.

Why it matters

BMPFast produces the nitrogen and phosphorus load calculations that engineers submit with stormwater permit applications in Florida. A permit reviewer, the engineer who signs the application, and the owner who builds the project all rely on those numbers. So accuracy is not checked once at release and then assumed: it is checked continually, by several independent methods, each catching a different kind of problem.

Validation

Are we calculating the right thing? The methods, equations and tables match the current permitting rules and the science behind them.

Verification

Are we calculating it right? The program carries out those methods correctly, for every kind of BMP and every combination of them.

The six methods

  1. Constant review of the requirements — the rules and the published methods behind every calculation.
  2. Feedback from permitting agencies — the people who review the permits.
  3. Feedback from users — the engineers who use BMPFast every day, and outside reviewers of its help.
  4. Automated testing — test suites that recalculate known projects and compare every result.
  5. Manual testing — every change is tried by hand in the program before it is accepted.
  6. Regular AI review — systematic reviews of the calculation code and the documentation.

The rules we follow are written down in the Testing Protocol for BMPFast (Wanielista and Eaglin). Each method below links to the parts of it that apply.

Item resolution and details are available to permitting agencies upon request — for any item raised in testing or review: the question, the decision, the change made and how it was verified. Email info@bmpfast.com.

1Constant review of the requirements

BMPFast follows Florida’s Environmental Resource Permit Applicant’s Handbook, Volume I (the June 28, 2024 edition), and the methods it relies on. When the rules change, the program changes. Its compliance report is laid out in the order and terms of Section 9 of the Handbook — the land-use and hydrology tables, the event mean concentrations, the loads, the required treatment and the treatment train — so a reviewer can follow each number back to the rule that calls for it.

Each calculation is tied to its published source. Examples:

  • buffers and filter strips follow Appendix O of the Handbook — the retention or detention calculation for the volume captured, then the removal for plants, soil and media as a treatment train (Harper and Baker, 2007);
  • the vegetated natural buffer removal tables are based on EPA/600/R-05/118 (Mayer et al., 2005);
  • vegetated filter strips with engineered media are supported by a peer-reviewed Florida field study (Shokri et al., Journal of Environmental Management, 2021).

Where a rule and the program disagree, or a limit is not supported by data, the program is changed to match. For example, the buffer tables have data for slopes of 2 % and 6 % only, so BMPFast now refuses a steeper slope and says why, rather than calculating outside the data.

2Feedback from permitting agencies

BMPFast comes from the team that developed BMPTrains for the Florida Department of Transportation. Its removal calculations and tables were developed and approved for use with the involvement of FDOT and water management district staff. Questions and comments from permit reviewers feed straight back into the program: when an agency asks how a number was reached, the answer becomes a clearer report, an equation shown on screen, or a correction.

Different review agencies accept or reject different BMPs. BMPFast’s aim is to let every BMP and every design be considered — and to show exactly how each one was calculated, so the reviewer can decide.

3Feedback from users

Engineers using BMPFast on real projects are the best test of all. Their questions and reports — sent to info@bmpfast.com, raised in the training course, or found while preparing a permit — are each looked into, answered, and where needed turned into a change.

The help pages are reviewed the same way. Every window in the program has its own help page on bmpfast.com, describing each input, result and button. Each page is checked against the program, then read by outside reviewers, and every question it raises is recorded and answered before the page is considered finished.

4Automated testing (test suites)

BMPFast has an automated test suite of more than fifty tests that runs against the calculation code. The heart of it is a set of worked example projects — retention, wet detention with harvesting, permeable pavement to detention, multiple BMPs in series, swales with groundwater discharge, multi-catchment networks, and examples from the BMPTrains training material. Each project is opened and recalculated, and every result is compared with its stored expected value. A change that moves any result, anywhere, is caught.

Other tests check the treatment-train routing (how catchments are grouped and calculated through each other), each analysis type’s targets, and the design curves that sweep a design value and plot the result.

5Manual testing

Every change follows the same path: it is specified, questions are answered, it is built and tested, its screens are reviewed for clarity, and then it is tried by hand in the program. Each change is tracked in a development plan, and is not marked complete until Dr. Ron Eaglin has used it and confirmed it works.

A hand test checks what the automated tests cannot: that a value typed on a worksheet reaches the calculation, that a report reads correctly, that a warning appears when it should — and that a worked example gives the same answer by hand as in the program.

6Regular AI review

The calculation code is reviewed regularly with AI tools, looking only for correctness: lookup and interpolation, the efficiency tables, and the special cases such as BMPs in series. Findings are confirmed by running the calculations themselves, not just by reading the code, and each is tracked until it is fixed.

AI review is also used to compare the help pages, word by word, against what the program actually does. That comparison has found labels, units and limits that disagreed with the code — each one recorded as a question, decided by the engineer responsible, and then corrected in the program, the help, or both.

AI is a reviewer, not a decision-maker. Every finding is checked, and every engineering decision is made by a person.

What you can check yourself

  • The Equations page. On every treatment worksheet, the Calculate button switches to an Equations page that shows the equations used and the values that went into them.
  • The reports. Each BMP report lists its inputs, the watershed values it used and the results; the compliance report follows the Applicant’s Handbook, Section 9.
  • Warnings. When a value is outside the range a method supports, BMPFast tells you on the worksheet, and the report notes it.
  • The help. The help page for each window gives its equations, limits and sources.

Found something that does not look right? Email info@bmpfast.com with the project file if you can. Every report is looked into.

Back to BMPFast