In this article I review how Danomics’ Flow tools for performing unit conversions and log splicing work together to ensure the best possible results. I’ll cover what the unit conversion flow tool does and why you should run it early in your data preparation workflow. Then I’ll explain how the log splicing tool works and what the outputs of it are. Finally I’ll review why it is important to run these tool in conjunction and what other tools should be considered before and after.
LogUnitsConversion Flow Tool
The LogUnitsConversion Flow tool performs three critical operations:
- Renames units mnemonic to a common units name
- Converts values to Danomics’ standard unit space
- Converts values to either Imperial or Metric units
Much like well logs, curve units often have their own set of mnemonics. For example, the bulk density curve, which is usually expressed in grams per cubic centimeter may have units such as G/C3, G/CC, G-CC, G/CM, and G/CM3. In this case, for consistency, it is desirable to rename all of the units to a common unit mnemonic such as G/C3. This will reduce problems that might arise when doing unit-dependent statistics later. Additionally, not all logs of the same type are expressed in the same unit. For example, porosity is commonly expressed in percent and decimal (each of which also have myriad unit mnemonics). In this case it is crucial that we convert the percent to decimal to avoid working in mixed units. Finally, logs can either be in Imperial or Metric units. Although Danomics can convert units on the fly, in some cases, such as when performing a final export, it may be desirable to convert units to one standard or the other.
LogSplicing Flow Tool
When you have more than one LAS file for a well, or when you have multiple curves of the same type in a LAS file, you will want to combine them to ensure complete data coverage. Danomics composites curves by default in the CPI on data import. What this means is that the first non-null value at each depth step is taken. This can be problematic if the curves are potentially of different scales due to tool differences or changes in the wellbore environment. To improve upon this we have introduced the LogSplicing Flow tool which has more sophisticated behavior.
The LogSplicing Flow tool does the following:
- It filters out log curves via statistical checks on their range +/- units
- It scores the log curves using a combination of their length and the number of internal nulls
- It selects the log curve of each type with the highest score and labels it as the primary curve
- It then (optionally) normalizes log curves when there are overlapping samples so that the data ranges match
The filtering and scoring are contained in the CPI Config and can be overridden as needed.
Unit Conversion and Log Splicing Together
The Log Splicing flow tool can be ran independently, however, performance will be optimized if ran in conjunction with the unit conversion tool. There are two reasons for this:
- By converting all of the units to a common unit space we will ensure that we are not trying to use normalization to stretch and squeeze mismatched units together
- Many wells may be missing units and the unit conversion tool can address this problem
An example flow is shown below:
In the LogInput tool it is important to select “Output all” as the option when splicing if you have multiple LAS files per well.
In the LogUnitsConversion tool you can select the operation you want to perform and the desired output units.
In the LogSplicing tool you can either have it splice curves of each alias type or specify which ones to operate on. You can also set if normalization will be used and how many overlapping samples are required for normalization. The Output mode lets you set if only the primary curve will be output (Splice) or if all curves will be output (QC).
Other Flow Tools to Consider
The splicing and unit conversion should typically be done after you have already done other data clean-up operations. For example, you may want to run the log and mnemonics health checks first, then FixLogDepthProblems, then trim flatspots off logs with the NullRepeatedLogSamples tool, and then run the Units Conversion and Log Splicing. For example, see the image below.