Tackling Large Projects: From 10 to 10,000 scans

How do we expand into large projects? This is a common question heard from providers of scanning services. Typically a firm has success in executing short projects lasting a week or so, and wants to tackle bigger projects of significantly longer duration. It would be nice if ramping up were a simple matter, but the reality is that scanning projects are complex, and that this complexity does not scale proportionally with project size. In this article we will examine some of the best practices and pitfalls in executing large projects.

Scan count. A good estimator of a projects size is the scan count. Often practitioners will use total number of points instead of the total number of scans, but for our purposes scans are better because they give a more realistic estimate of complexity. More points do require more time in the field and typically yield more detail, but its the number of scans that causes most headaches. I typically recommend that scans be collected at consistent resolution, and that you choose the highest practical resolution available (within reason.) Data storage is cheap and getting cheaper, and the extra few moments in the field have a low marginal cost and will eliminate expensive return trips to the site.

Field personnel. Large projects often require multiple crews. If so, make sure to they all have consistent training and follow written Standard Operating Procedures (SOPs). Often the field time is the largest controllable expense, so be sure that the crews are being efficient, but not sloppy. For a small project it may not be a big deal to return to the scene and retake a scan, but this can be difficult and expensive to do on larger projects, especially those executed over a short window of opportunity. Inasmuch as possible, the SOP should ensure that crews leave the field with all the data that could be needed, so there will not be any surprises back at the office, or at the clients. One important job of the field crew is to ensure accuracy of the registration network. Ideally, the registration should be complete before the crew leaves the sitefinding a blunder in the office a few days later can be very expensive. This leads to the next topic: registration.

Registration. For few scans, one can register by hand or use commercial cloud-to-cloud registration software. Eyeballing the results may be good enough. Clearly this approach does not scale effectively. Furthermore, scanners may not have sufficient accuracy to hold tolerances over longer distances, so traditional surveying is often incorporated. Registration errors may require costly return trips to the field to fix, and will delay the project. The later the problem is found, the more effort is wasted; costs can skyrocket, and the client relationship is strained. Confidence in the registration of hundreds or thousands of scans coupled with survey control is gained by implementing operationally and mathematically rigorous procedures that we cant get into here. The upshot is that registration is a tricky, complicated matter, and one of critical importance. If your business doesnt have an expert on staff, be sure to work with a partner or consultant to ensure that you get this right. The penalty for not doing so can be very high.

Processing. Once the registration and initial preparations are complete, measurements, models, reports and the like can be produced. It is important that this processing is done both consistently and traceably. The basic idea is to minimize risk through staged results and continual quality checks.

Consider the example of a piping designer routing a single new line to connect two tie points. Rather than do it all at once, it is prudent to separate the process into several steps: determine the coordinates and uncertainties of each tie point, route the line, then create a clash report to verify that the new design does not interfere with existing hardware.

At each step the information is be recorded, along with meta-information such as date/time, operator, supervisor approval, etc. Later on, a quality operator checks the pieces separately. If an issue does arise, it is much easier to resolve since it is contained within a limited scope. Commercial software can help by providing automated tools that remove some of the guesswork from routine tasks and also maintain an audit trailparticularly useful for clash reports that must be signed off by a supervisor.

It is important to note two caveats: no software is perfect, so there will always be a need for a quality control step (some vendors claims for completely automatic processing notwithstanding!); and current products tend to focus on one particular application, meaning a package designed for industrial plants may not be useful for, say, architectural applications.

In summary, the key to successfully executing large projects is to have a consistent, predictable, and traceable straight-line process. Loopssuch as return trips to the field, or tweaking registration after extracting measurements from the dataare productivity killers. Your SOP should eliminate or defer as much decision-making as possible: collect as much data as you can, then youll have it later if you need it. Build quality control and assurance into your SOP at multiple steps to nip issues in the bud. Often the clients business processes drive the need for consistency and reporting, so be sure to understand their requirements thoroughly before tackling that big project. And most of all remember that ramping up takes patience and the wisdom that comes with experience: be sure to take it slowly.