StatCalc

Total Inputs: 0

LSL

USL

Mean

Median

Range

6σ

−3σ

+3σ

Input #1

Distribution Type

Click to Change

Delete this input?

Standard four-function calculator: +, −, x, ÷, =.
√x, x², 1/x, and % apply to the number currently on screen.
MS stores the current value in memory, MR recalls it, M+ / M− add or subtract from memory, MC clears memory.
C clears the whole calculation; CE clears just the current entry.

√x²+y² and (√x²+y²) x n (the Fudge factor) are mode buttons — they select what RMS = will do, but don't calculate anything by themselves.
Press √x²+y² to start accumulating. Type a value, then press + (or the same button again) to square it and add it to the running sum. Repeat for each value.
Press RMS = to finish: it squares whatever's currently entered, folds it into the sum, and takes the square root — so you don't have to press + before the last value.
The display keeps the running list of values, the sum of squares, and (once a calculation completes) the square root all visible together.
(√x²+y²) x n (Fudge factor) only lights up once a calculation is complete. Press it, type a multiplier, then press RMS = to multiply the completed result by that factor.
C clears the current calculation but stays in whichever mode was active, ready for re-entry. CE clears just what's currently typed, without touching the running sum.
√x, x², 1/x, % transform the entry — the result still gets squared like any other raw number when + or RMS = is pressed.

Results show Mean, Median, Range, 6σ, −3σ, +3σ of the simulated outcome, plus σ and % out of spec below.
Requirements (Nominal, Lower Limit, Upper Limit) describe the overall simulation — Lower/Upper Limit are used to report the % of simulated results that fall out of spec.
Enter Inputs (or Edit Inputs, once you've saved at least one) opens a separate input screen, where you define each item in your simulation.
- Each input needs one of three combinations: USL & LSL, or Mean & 6σ, or Mean & −3σ & +3σ. The rest fill in automatically once you press Enter Input.
- USL/LSL are treated as independent reference values when you use Mean/6σ or ±3σ instead — they're never overwritten by the math.
- Use the ◀ ▶ arrows to page between inputs. New Input creates a fresh one (once the current one is saved); Delete Input removes whichever one is on screen, with a confirmation first.
- The number pad on this screen only enters values — there are no mathematical operations available, since each field just holds a single number.
- C only works on a field that's currently selected — tap into a field first before pressing C.
Sample Size sets how many trials the simulation runs (10,000 by default).
New Sim (appears after a simulation has run) clears everything — inputs and results — for a completely fresh simulation. It asks for confirmation first, since it can't be undone.
Run Sim draws random values from each saved input's distribution, sums them, and repeats for the sample size you set.

About StatCalc.online — and a Brief History of Tolerance Simulation

Dimensional Management was started in the 1980s at GM by Bob Kaphengst on their truck programs. John Mathieson joined them, then moved over to the CPC division. Curt Larson was hired by Bob at Truck & Bus, then joined John at CPC. John and Bob left GM and formed the pre-cursor to Dimensional Control Systems with Tony Cross. When Bob, John and Tony started their own company, Curt joined them.

After the passing of Tony Cross, John and Bob created Dimensional Control Systems, and Curt joined them leading their Sales & Marketing efforts, special engineering projects (can you say Ferrari F360 Modena?) and software development.

Dimensional Management is the technology of defining and reducing dimensional variation of assembled products, developed specifically for the auto industry, with GM and Chrysler the early adopters of the methodology pioneered by the founders of DCS.

In those days, manufacturing engineers did rough tolerance calculations using RMS — Root Mean Square, with the calculation of the “square root of the sum of the squares”: a down and dirty method to determine the accumulation of variation. When building 100,000 vehicles (or more) a year, with each part having a range of variation, what would be the resulting range of variation? To make the numbers work out, they often multiplied the RMS result by a “fudge factor,” usually 1.5, to have a working number to see if they were in the ballpark.

With the advent of computer-aided design and computer-aided manufacturing, where hand-built models became a thing of the past and coordinate measuring machines became the norm to measure part variations to hundredths and thousandths of an inch, the industry was ripe for a new technology to manage and control variation.

Early truck program dimensional management review

While a competitor had a tolerance simulation software package, it was originally on a mainframe and users logged in with their computer and wrote code to simulate the accumulation of variation in a 3D world. They eventually developed a pc-based program, but it remained solely a code-based application, requiring specialized engineers/programmers to write the simulation code for each vehicle, part and assembly.

Enter Dimensional Control Systems and their creation of 3DCS — a graphics-based tolerance simulation program, with Ying Qing Zhou the brilliant lead developer who wrote the initial code. 3DCS downloaded the CAD data, performed Monte Carlo simulations on part tolerances, and graphically displayed the variations. Users could zoom in and see the impact of tolerance accumulations over thousands of samples.

Curt Larson was given the job of leading the development team, provided the logics to the developers to code, plus worked out the menu systems and UI layout of the first version of 3DCS. Also at this time a version that integrated with AutoCAD was developed that Curt was responsible for.

At the same time, Curt designed and DCS built 1DCS — a product that was an improvement over the RMS method for simple calculations that broke down tolerances from an objective to sub-assembly and piece-part levels, as well as simulated tolerances from the bottom up using Monte Carlo simulation.

Curt also designed a statistical calculator that was never built.

Original never-built statistical calculator design

Fast forward to the world of AI. As an exercise to research AI, and to fulfill a long-held dream to have a statistical calculator that would perform tolerance simulations, Curt took his initial design and updated it to modern design practices and to fit on a phone, then worked with Claude to build a statistical calculator.

The result: StatCalc.online, a statistical calculator that will do Monte Carlo simulations on a number of variations on a mobile device.

Using StatCalc.online, Dimensional Engineers will be able to perform rapid statistical calculations of variations in a few short minutes to give better insight at the planning, styling and engineering phases of a program. While it does not take into account the 3D effects of variation, an engineer can get a better approximation of accumulated variation on assembled products, and use it to make better decisions early in a program, before full 3D simulations can be run.

The first release of StatCalc.online uses a standard distribution curve. The ability to change from a normal distribution curve to a skewed, bi-modal, flat and other distribution types for each tolerance has already been worked out. Additionally, the ability to Save and Open previous calculations is also being considered.

Dimensional Engineers: StatCalc.online is for you.

And for anyone who needs to rapidly perform statistical calculations.

As for Curt, he continues to drive innovation and technology, and develop new software apps that move technology forward.

curtlarsonpm.wordpress.com

STATCALC

STATCALC

ENTER INPUTS

STATCALC

HELP

STATCALC

ABOUT