Laboratory-Based Evaluation of Gas Well Deliverability Loss Caused by Water Blocking

Abstract

Summary Water blocking caused by invasion of completion fluids has been suspected to reduce gas well deliverability.1–5 However, this effect has not been quantified. We report results of a laboratory program to measure the water-blocking effect in core samples from a gas field. These data were mapped to a wellbore model to make deliverability predictions. The laboratory data consist of gas flow rate as a function of injected gas pore volume for various liquids (brine, methanol, toluene, isopropyl alcohol, and brine-methanol mixtures) at two saturation states (fully saturated with liquid, and containing liquid and trapped gas). We injected over 10,000 PV of gas in each test to mimic near-wellbore conditions. The data showed that the liquid displacement regime was followed by a mass transfer regime. The wellbore model had a time varying skin to account for the cleanup of the fluid invaded ("water-blocked") region. Cleanup occurs as gas flows past this high liquid-saturated region and removes liquid by displacement and mass transfer. We used the laboratory data to relate the reduced permeability of this region to pore volumes of gas throughput. We find that any loss in gas well deliverability recovers in two phases. The first phase corresponds to fluid displacement ("flowback period") and lasts for a few days at most. The second phase is slower and can last several months. Here, evaporation causes the deliverability to slowly increase. It is in this regime that adding volatile fluids, such as methanol, to the completion brines has advantages. Introduction Poor gas flow performance following well operations such as completions and workovers was recently observed in some wells in a gas field. Loss of aqueous fluids during these operations causes a ring of high water saturation around the wellbore. This can potentially reduce gas flow into the well, and this phenomenon is called "water blocking." The objectives of our work were to assess the impact of water blocking on well deliverability and to evaluate remediation possibilities. Water blocking has been suspected to reduce deliverability of gas reservoirs.1–5 Bennion et al.1 and Cimolai et al.2 assert that water blocking is a problem in which the in-situ water saturation is significantly less than "irreducible" water saturation. They present two field case studies to advocate their claims. The first case study is on the Paddy formation in Central Alberta [k~100 md; f=15%; Sw (in situ)=17%; Swirr (lab) = 43%]. The second is on the Cadomin formation in Alberta [k~1 md; f=5%; Sw (in situ)=20%; Swirr(lab)=52%]. Metheven3 discusses the performance of gas wells in the Frio and Wilcox formations in Texas. His data show that oil-based drilling fluids lead to significant improvements in gas productivity compared to water-based muds or invert emulsions. Laboratory tests indicate return permeability to gas after exposure to muds to be higher for oil-based mud than for water-based mud. Metheven suggests water blocking and vaporization of oil base filtrate by gas production as reasons for these differences at the laboratory as well as field scale. Holditch4 presents a numerical study of formation damage around a hydraulic fracture in a tight gas sand reservoir. He makes an interesting observation that formation damage can increase the capillary pressure of a rock and that this synergetic effect could lead to waterblock problems. Abrams and Vinegar5 use Computed Tomography to image the flow of nitrogen and brine in microdarcy gas sand cores. They claim that water block is unimportant if the drawdown pressure gradient in the region near a hydraulic fracture is on the order of several hundred psi/in. Stimulation using alcohol or surfactants did not significantly improve gas flow in these cores. On the other hand, MacLeod6 claims aqueous stimulation fluids containing alcohol have proven to be highly successful in stimulating gas production from problem wells in sandstone formations. The published literature does not contain a systematic set of laboratory measurements using different fluids and saturation states. There are no data on corefloods exposed to the large pore volumes of gas flow as would be expected in the near-wellbore environment. Also, laboratory data have never been mapped to a wellbore model to evaluate whether the effects seen in the laboratory are important in reducing well deliverability. Our study attempts to address some of these shortcomings. Approach Our approach consists of the following steps:Laboratory gas floods are conducted at ambient conditions to generate Krg vs. PVgas data.The Krg-PVgas curves for reservoir conditions are computed from the laboratory data.A well flow model with a time varying skin to represent the water-blocking region is developed. The permeability of this skin changes as gas flows past it, and is given by the Krg-PVgas curves.Well deliverability calculations are made for different conditions. Laboratory Gas Floods We conducted laboratory experiments on a preserved, composite (three plugs) sandstone sample with the following properties: f=16%, k=14 md, Swi=28%, L=16 cm, PV=30 ml. The experiments consisted of a series of room-condition, constant pressure- drop humidified methane floods of the core sample containing various liquids (brine, methanol, toluene, isopropyl alcohol, and brine-methanol mixtures) at two saturation states: fully saturated with liquid, and containing liquid and trapped gas. Brine represents the primary fluid in the water-block region; brinemethanol mixtures represent liquids used for remediation; methanol, toluene, and isopropyl alcohol are used as model liquids to help interpret the data. The saturation states represent the potential conditions in the near-wellbore environment. We present only selected aspects of the laboratory work in this paper, and the laboratory data and analyses are discussed in detail in Ref. 7. We also note that toluene was used as a model solvent to validate laboratory data and has no practical significance. Isopropyl alcohol was used as it has been used for stimulation purposes,6 and it provides a reference value for another potential remediation fluid.

Imbibition and Water Blockage In Unconventional Reservoirs: Well-Management Implications During Flowback and Early Production

“…For instance, the studies by Chowdiah (1987) and Shanley et al (2004) suggest that gas production suffers dramatically when water saturation exceeds 40-50%. A recent study involving porelevel modeling by Le et al (2012) suggested that displacement triggers rapid recovery of the gas relative permeability because capillarity plays a minor role. Beyond the studies performed at the pore level, analytical modeling has been attempted to understand the flowback data (Ilk et al 2010;Clarkson 2012).…”

Drying Rate Measurements in Convection- and Diffusion-Driven Conditions on a Shaly Sandstone Using Nuclear Magnetic Resonance

“…Drying driven by gas flow has only been poorly investigated. Some study has been done in the context of gas storage and production industry because of water blocking problems (Kamath and Laroche 2003;Parekh and Sharma 2007). Water blocking can appear for gas wells after injection of aqueous fluids during operations such as completion or workovers.…”

Enhancing Hydrocarbon Permeability After Hydraulic Fracturing: Laboratory Evaluations of Shut-ins and Surfactant Additives

“…Water-based fracturing fluids are commonly used to fracture tight reservoirs in North America (Wasylishen and Fulton, 2012), and it has been reported that only 5% to 50% of the injected fluid is typically recovered as "flowback" (Zelenev and Ellena, 2009). Water loss to the formation can have detrimental effects on the hydrocarbon flow from chemical and mechanical damage, for example clay swelling or rock softening (Alramahi and Sundberg, 2012;Das et al, 2014;Zhang et al, 2015), to relative permeability damage due to capillary trapping of the water phase and water blocking (Abrams and Vinegar, 1985;Bennion et al, 2000;Dutta et al, 2014;Kamath and Laroche, 2003;Mahadevan and Sharma, 2005;Liang et al, Submitted).…”