Controls of Wellbore Flow Regimes on Pump Effluent Composition

Abstract

Where well water and formation water are compositionally different or heterogeneous, pump effluent composition will vary due to partial mixing and transport induced by pumping. Investigating influences of purging and sampling methodology on composition variability requires quantification of wellbore flow regimes and mixing. As a basis for this quantification, analytical models simulating Poiseuille flow were developed to calculate flow paths and travel times. Finite element modeling was used to incorporate influences of mixing. Parabolic velocity distributions within the screened interval accelerate with cumulative inflow approaching the pump intake while an annulus of inflowing formation water contracts uniformly to displace an axial cylinder of pre-pumping well water as pumping proceeds. Increased dispersive mixing forms a more diffuse formation water annulus and the contribution of formation water to pump effluent increases more rapidly. Models incorporating viscous flow and diffusion scale mixing show that initially pump effluent is predominantly pre-pumping well water and compositions vary most rapidly. After two screen volumes of pumping, 94% of pump effluent is inflowing formation water. Where the composition of formation water and pre-pumping well water are likely to be similar, pump effluent compositions will not vary significantly and may be collected during early purging or with passive sampling. However, where these compositions are expected to be considerably different or heterogeneous, compositions would be most variable during early pumping, that is, when samples are collected during low-flow sampling. Purging of two screen volumes would be required to stabilize the content and collect a sample consisting of 94% formation water.

Monitoring well utility in a heterogeneous DNAPL source zone area: Insights from proximal multilevel sampler wells and sampling capture-zone modelling

“…The objective being to ensure sampled water is in quasi-steady state interaction with the formation. Such volumes are consistent with the recommendations of Martin-Hayden et al (2014) who illustrate a 'partial purge' condition may persist over c.3 SV. In passing, we concur with the recommendation of Martin-Hayden et al and Robin and Gillham (1987) that where the pump intake is positioned within the well screen and there is insignificant drawdown, screen volumes should be used as the unit-of-measure for quantification of purging as they are technically more defensible than well volumes and constant with time.…”

“…It can hence be reasonably inferred monitoring well sample concentrations obtained during low-flow purging (and indicator chemical parameter stabilization) that involves removal of low volumes (around 1 SV or less), will be very sensitive to pump intake position, especially where marked variability of concentrations occurs with depth; a default expectation, and hence issue, for DNAPL source-zone areas. Concentration sensitivity over < 1 SV purged (but extending up to around 2 SV or more) is similarly predicted by Martin-Hayden et al (2014) even under homogeneous aquifer conditions alongside secondary sensitivity to the wellbore flow representation modelled.…”

“…Purged water was monitored for stabilization of water quality parameters with removal of a minimum of 3 well volumes (24 l, assuming typical rest water levels at 2 m bgl) prior to taking samples for VOC (volatile organic compound) and supporting hydrochemical analysis. Similar to Martin-Hayden et al (2014), we later report and simulate time-variant data as screen volumes (SV) removed (purged) as opposed to well volumes. For a 3 m × 50 mm diameter well screen, 3 SV is equivalent to 18 l. With minimal drawdown sampling where the pump intake is located within the well screen, little sample water is expected to originate from the blank-casing.…”

“…Prior studies suggest purging of 3 SV is likely sufficient to obtain a quasi-steady state sample from a monitoring well; for example, the modelling of Martin-Hayden et al (2014) to account for partial mixing and transport times to the pump, and the modelling of McMillan et al (2014) to account for the delayed arrival of casing water at the pump. Our predictive model simulations support these prior conclusions.…”

“…Whether purge times (and hence purged volumes) are adequate to attain a steady-state FWA sample concentration, however, is a key issue (Martin-Hayden, 2000;Martin-Hayden et al, 2014;McMillan et al, 2014). The variation in delayed arrival times within the sample taken of groundwater entering the monitoring well screen at distance from the pump intake becomes of critical significance here .…”