PostFrame Manager Help

Three Random Buildings Test

Overview

Unlike creating live buildings to see what works, and unlike painstaking work with tablet, calculator and pen, this is software. Virtual buildings. Although this software can reasonably reproduce what you may have built in a price book, it costs you NOTHING to change things and experiment. The problem is that until now there's never been software this fast and easy, so few people have had any way to freely experiment. Times have changed.

We have been in the fortunate position of working with a number of builders and have seen a number of real live designs created with this software. Every time we get hold of a live design, we use this to test in the lab, where the development team has a copy of software that iterates through a number of different tweaks to the key selections; the resulting building price is recorded for every iteration. Our head of sales has mentioned this in breakout sessions at trade shows, we refer to this as the A.I. (artificial intelligence) copy. The idea was and is to let the software automatically derive the least expensive price regardless of the builder assumptions. Using software this way is called Virtual Construction.

Since there's never been a commonly accepted view of Virtual Construction, there's also no common way for anyone to examine assumptions or methods. This page details how to do this.

Buildings at Random

The idea here is reasonably simple: draw 3 random building sizes and save each one. By Random we mean lengths that aren't a bay size multiple. As in 53', 44' and 61' buildings. Or 37'. Or 33' 6". Then, just to keep things purely random, add some openings to each one. Don't add the same 8'x12' overhead in every endwall. Random openings in Random walls. It's OK if a Random building is an ideal bay space. They don't all have to be odd spacings. But they all shouldn't be ideal multiples, either.

Take each building in turn, and apply your control to it. Let's say that you normally do 8' spacings and you want to see the effects of Post Spacing Method. That means that all 3 buildings will get the same 8' spacing set up, and then you vary the Spacing Method. Record the resulting price for all 3. Change the Spacing Method. Repeat.

If your control is lumber lengths, try that as well. Again, assume you normally use 8' spacing and your contention is that your lumber needs to be 8' and 16' only. In that case you restrict the lumber on the first run. Record the prices. Then allow more lengths to appear. Record the new prices.

What To Make of the Results

Generally what you ought to find is that, on average, letting the software default to even post spacings and using Most lumber lengths rather than restrictions will result in the best prices. Again, this is On Average.

So what does this mean, overall? In part this depends on your philosophical approach. There's no doubt that fine tuning a given building by hand can and may result in some savings. The amount is based on what was changed and how. In part this also depends on materials pricing and having an accurate labor pricing model. The question is how much savings do you get for fine tuning by hand. You can't know if you don't look for it.

The Argument For Virtual Construction

If you implement testing of random building sizes as advocated, undoubtedly you will run into interesting data. Perhaps you'll find that using spacings of 8' on sizes above a certain benchmark and that 6' spacings below the benchmark size is less expensive. Maybe not. Perhaps you'll discover that Spacing Method is less important when you have a number of openings than when you don't. We can't say. But what we can say is that you are never going to really know -- you may suspect, but you won't know -- until you try a handful of random buildings, vary the settings, and see what the effects actually are.

You may have looked at this product as a way to implement your price book, and if it didn't fail, maybe use it to do some small variations or perhaps, with time, do some more designs. But construction is highly variable, and your experience says that software can't really do the job. But... think about this a different way. This is software. In concept it's no different than a spreadsheet: change this data point and recalculate to see the effect. It costs you nothing other than a minute or two of your time to use it this way. You can test any building easily and recalculate the effects. Just like a spreadsheet.