Integrating AI and Monitor While Drilling Systems

One of the biggest opportunities in AI may be in an unexpected place - Specialty Geotechnical Construction. 

Currently, the industry relies on sophisticated Monitor While Drilling (MWD) data collection systems. These systems attach to various parts of drill rigs, pumps, cranes and other items used in geotechnical construction. While attached, these systems collect a variety of data points including torque, rotary pressure, grout flow rates, drill times and much, much more. 

This data has been used almost exclusively as quality control. Most times, if not all, the work in the ground will never be able to be seen and inspected. Therefore, there needs to be a way to verify that the work installed has been installed per specifications. 

Other companies may go a step further and use the data for more sophisticated past cost estimating data, however for the most part, the data is collected, submitted to the owner, and then forgotten about forever. 

What if there was a better use for this data? What if there was a solution to increase installation speed, decrease quality mistakes, and reduce operator training times?

The solution proposed is creating an Artificial Intelligence model that could integrate with a MWD system. The appeal is two folded. First, the system would be able to maximize a certain parameter of choice. The easiest example would be a company wanting to maximize penetration rate so a system is designed to give an operator tips on certain parameters given the site’s geology and past data learned on the site. Second, the system would help to speed up operator training. Obviously, the industry still needs skilled operators, but having an onboard system that could be taught by a more experienced driller could then be transferred to a second rig where a less experienced driller is operating. The system could then give the less experienced operator tips on how to better install piles based on the data recordings from the more experienced operator. 

The system would not just be limited to a “co-pilot” training method where training data is derived from recording operators. The system could be trained in a variety of ways both with previous data sets but also with empirical relationships that have been verified either in public research or with in house test programs. For example, a jet grout test column would be a perfect data set for an AI to make future predictions as the algorithm would have all relevant grouting parameters used in creating the jet grout column. Not only the grout parameters, but the drilling parameters as well as the site geology of what was drilled and grouted through. Based on these data points as well as the field measured Jet Grout column diameter, the algorithm could make predictions based on changes in grout parameters or predicted soil conditions based on drilling parameters. This is essentially what a test program does anyhow, adding artificial intelligence simply helps make better educated guesses of what is happening DURING construction instead of waiting until AFTER. 

This is also not a totally new idea. Plenty of publicly available research has looked into the correlation and prediction of jet grout column diameters based on known data sets. These research papers use Support Vector Machines to create Jet Grout diameter predictions and general diameter guidelines for pre-construction planning. 

Another example of the successful use of AI integration in MWD systems is in mine exploration. Exploration miners have already successfully used AI to identify the soil classification during drilling with roughly 90% accuracy. 

Obviously, this system would be far more complex than it sounds, with many other variables to take into effect such as tool condition and choices which are much less easily quantifiable. However, the current AI platforms are providing the bare bones infrastructure needed to create case specific Artificial Intelligence. Firms that understand the tools at their disposal and how to leverage them will gain an unfair advantage in the marketplace. Deep Foundation and Ground Improvement Construction is a great place to start as these industries are built on huge data sets.