Detecting the presence of two hydrogen bearing fluids through the T2 distribution under circumstances similar to those found in near surface NMR

Abstract

The use of low field nuclear magnetic resonance is becoming more common as equipment becomes more advanced. There is a rising interest in using low field NMR for various applications like identifying the presence of multiple hydrogen bearing fluids in the near surface. A common way to detect multiple fluid types using NMR is by resolving multiple peaks in the T2 distribution. Certain types of sediments that we might find in the near surface make it difficult to do this. In higher field NMR it is easy to manipulate measurements to suppress the wetting fluid signal which manifests as a separate peak shifted to a lower T2 time. We can then interpret this peak separation to indicate the presence of multiple hydrogen bearing fluids. One way of manipulating the measurements is to perform the enhanced diffusion method (EDM). It has been suggested that this should be performed with the highest possible gradient, i.e. higher field, to maximize the T2 contrast. However, this has not been explored in lower field settings. Here we begin to look at some of the practical aspects of performing EDM in lower field NMR. We anticipate that performing EDM in a low field will cause a significant decrease in data points. This leads to a loss in resolution in the recovered T2 distribution so that even though we are efficiently shifting a portion of the recovered T2 we cannot tell when we look at the T2 distribution. For this reason a large part of this work is implementing a Tikhonov inversion algorithm to gain back some resolution in the T2 distribution. We implement the logarithmic barrier method to enforce boundary constraints on the T2 amplitude, and then pick the regularization parameter using the generalized cross validation (GCV) method. In our experience other methods for picking the tradeoff parameter result in over-regularization. We then test this algorithm on EDM data produced using a low field 275 kHz laboratory NMR system. Prior information about the samples used gives us certain expectations about the amplitude and placement of peaks in the T2 distribution. The result is that we are able to recover expected bimodal T2 distributions. Overall, we gain an insight into some of the practical aspects of performing EDM in a low field setting. The inversion algorithm proves to be more than adequate in resolving the T2 distribution in a low field setting and can be used in a high field setting with perhaps even more success. The results here can be extended to use with lower field near surface borehole equipment and with advancements in surface NMR equipment we should be able to apply these concepts to surface NMR measurements.