86
A similar comparison was done for a sample collected from the field
study using the measured
pressure and molar gas composition with the comparison shown in Figure 58.
Figure 58: Comparison of the measured (left) and calculated (right) composition of gas absorbed in the lubricant at
the discharge manifold. Note: Scale is logarithmic to show traces of heavier hydrocarbons.
In contrast, the measurements shown in Figure 58 show a lower concentration
of methane, a
higher concentration of ethane, and a lower concentration of the heavier hydrocarbons than
expected. The gas molar composition for this field visit was 96.6% methane, 3.3% ethane, 0.1%
propane, 0.011% butanes, and 0.002% pentanes. Large variations were observed between the
measured and expected values for both the experimental results and the field study. Thus, the
degassed lubricant was collected as shown in Figure 59 and analyzed for residual gas species.
88
The density of the degassed lubricant was measured with an Anton-Paar SVM 3000
Viscometer-Densitometer with an uncertainty of 0.00005 g/cm³ for density measurements. This
was done at temperatures from 20-80°C (68-176°F) with the results shown in Figure 60.
Looking at Figure 60 in detail, one notes that the used lubricant had a density that was slightly
lower than the neat lubricant up to a temperature of 60°C. At temperatures above 60°C, the
density began to diverge from the linearity seen at lower temperatures. Upon inspection, it was
noted that bubbles were forming in the densitometer at temperatures above 60°C indicating that
there was some gas still dissolved in the lubricant even after the degassing process. This
indicates that this method for determining solubility may not be ideal and the author points to the
use of viscosity mixing rules and the data from Seeton (2019) as a better source as described in
section 3.5 - Viscosity - Comparison with Previous Work. Following is a
list of potential sources
of error in the current study and potential mitigation techniques for future studies.
1) Hydrocarbon gases are still entrained in the lubricant sample after pre-heating and
depressurization into an evacuated chamber. Future researchers should
identify better
separation techniques than those listed here.
2) The sampling process allowed partial flashing of the samples allowing some dissolved
gas to come out of solution. This could be mitigated by using a sampling cylinder with
back-pressure regulation capabilities.
89
Chapter 5
–
Modeling Compressor Lubrication
5.1
–
Model Purpose and Description
The previous sections have focused on measuring and estimating the viscosity of a lubricant at
conditions relevant to reciprocating natural gas compressors. However, for the viscosity to be
finally applicable
to a compressor, a determination of how the lubricant protects the moving
parts
–
specifically the piston rings and compressor cylinder needs to be investigated. Film
thicknesses have been extensively measured and modeled for internal combustion engines, but
relatively little work has been done for reciprocating gas compressors. As mentioned previously,
a compressor’s piston rings have the added complicati
on of sealing between high pressure
gradients, often higher than those seen in internal combustion engines. This provided the
impetus to model the lubricant film on the compressor cylinder to
determine both the volume
and viscosity of lubricant necessary to adequately lubricate the compressor at different
operating conditions.
The modeled system contains two identical piston rings preventing contact between the piston
and compressor cylinder as depicted in Figure 61.