Windbreak Design

Many homes are perched on high slopes that give them spectacular views and gale-force winds. Enjoying the view is only possible from indoors, and beautiful patios and porches go unused.

One solution to this problem is to build a wall. This solution gives the home-owner the worst of all possible worlds: a glass wall is still a wall, and the open freedom of being outdoors is lost. As well as being inherently unesthetic, glass walls require cleaning and maintenance, and are a hazard to wild-life, especially birds.

A glassed-in patio has all the limitations of a sun-room with none of the all-weather value. They are one of those solutions that are "simple, straightforward, and wrong." But what are the alternatives, and how to you choose one without a lot of expensive trial-and-error?

Simulation

Computational physics is the science (and art) of modeling the physical world with computers. One of its major areas of application is the design of air-craft, which involves simulation of air flow over the wing. Windbreak design is a more modest application of computational physics, but it illustrates many of the strengths and limitations of the approach.

Simulation involves two basic aspects: computer software that models the physics of the system in question, and some kind of data description language that lets the user tell the software what it should be modeling.

In the present case, the software used is a system that solves the Navier-Stokes equations, which describe fluid flows. The data description language is a series of files that contain the geometry we want to model. The most important of these are just image files, so we can add and subtract elements from the geometry using any standard image-editing software. This makes trying out different geometries very easy.

Initial Situation

The house in question has a broad patio with a short, 2 foot tall and wide planter at the end. Beyond the patio the slope drops off very steeply. A profile of the geometry is shown below:

The stick figure standing on the patio is 5 feet 8 inches tall, which should give a sense of the overall scale.

Because the software used is two-dimensional, this profile is all that it knows about. Doing full three-dimensional simulation on a desktop computer is still a very challenging problem, as it requires thousands of times the speed and memory of two-dimensional simulation.

Modeling a modest breeze in the original geometry shows the problem clearly. In the following false-color images the red areas are high wind velocity, green is moderate and blue is low. As can be clearly seen, a person standing on the patio would be exposed to a higher-than normal wind velocity: the wind hits the slope, turns up it, and flows over the planter and across the patio. Near the ground behind the planter the wind speed is low, but anyone sitting on the patio would be exposed to high winds (the bright green band) right near their head.

Adding a wall to the top of the planter improves things in some ways, but not in others:

As can be seen, the wind tends to curl above the patio, driving a "tread-mill" swirl of wind near the ground, shown in the false-color image as a green lozenge near the stick-figure's feet, and extending up to almost shoulder-level. The simulation doesn't capture all the nuances of fluid flow, but the strength of the swirling seen here strongly suggests that the air behind the barrier will not exactly be still.

This is an important first lesson about fighting the forces of nature: trying to stop them is rarely the right thing to do. No matter how much force we can bring to bear on the problem, nature can bring more. The thing we have going for us that nature doesn't is brains, and we need to use them to persuade rather than to bludgeon.

This result also helps us set reasonable expectations: if you want perfectly still air, stay inside. Even a solid barrier is going to have some wind behind it.

Alternatives

Rather than trying to stop the wind, suppose we try to redirect it?

This can be done in a variety of ways, but looking at these preliminary results, and some vector flow diagrams not shown here, suggests that lifting the wind downslope may help keep the patio calm. Notice also how the house itself creates still air in front of it by lifting the wind over it. This effect will also play a role in designing a more gentle solution to the problem.

Placing a brush hedge downslope does indeed lift the wind off the front off the patio near the slope:

This has a couple of undesirable consequences, the most important of which is the lowering of the air-stream near the house, so people standing on the patio will get a stronger wind there than before. It seems likely that it is possible to do better than this with relatively small modifications. Notice, however, that this is already an improvement over the original situation, is lower maintenance than a glass wall, and preserves the open, uncluttered view.

Because simulation is so cheap, it's easy to try different placements of trees to improve things. As noted above, the house itself lifts the airflow upstream from itself. This is a common phenomenon in areas of physics like fluid mechanics where the behavior of a system is governed by non-linear differential equations. The conditions at far-distant boundaries can have a large effect on events that naively we might think are unrelated.

Adding a tree to the simulation adjacent to the planter has the effect shown below:

The tree model is fairly crude. Unlike the hedge downslope, which is assumed to be a solid barrier, the tree is modeled as something fairly open. Because the trunk is small, even it is has had gaps left in it to let the wind flow past it. Otherwise, in this two-dimensional simulation, the model will think that even a very thin barrier is a solid block to the wind.

The tree is about eight feet tall and has a fairly bushy crown extending from four feet off the ground upwards. Unsurprisingly, given its similar placement and structure, its effect on the flow is similar to that of the solid windbreak modeled initially: it lifts the wind up while at the same time driving a "tread-mill" at ground level that indicates strong breezes will still be experienced by someone sitting on the patio.

To combat this effect various other geometries were investigated, the most successful of which is shown here:

By moving the tree toward the center of the patio it acts as a prop to keep the wind high above the ground, leaving the area beneath relatively still.

Conclusion

This has been a simple over-view of the use of computational physics in an ordinary landscaping problem. It deals with only one aspect of the problem: the placement of hedges and trees to maximally reduce the wind, and shows that carefully chosen natural features that redirect the wind are capable of doing at least as good a job as a solid barrier.

The map, as they say in geophysics, is not the territory, and a simulation is not reality, particularly when it is a two-dimensional non-turbulent simulation of a three-dimensional turbulent reality. But with cautious and expert interpretation even a very limited simulation can provide some guidance. The images shown here are fairly crude approximations to the reality they model, but an intelligent designer can use them as a useful and valuable guide.