A Laboratory Study With Field Data of Downhole Gas Separators

Abstract

Summary Downhole gas separators are often the most inefficient part of a sucker-rodpump system. This paper presents laboratory data on the performance of fivedifferent gas-separator designs. Only continuous flow was studied. Field dataare presented on two of the designs. The field data indicate that success orfailure of the gas separator is dependent upon the fluids and wellborepressures as well as the mechanical design of the gas separator. Successful andunsuccessful examples of gas-separator performance in the field are shown alongwith field fluid data properties. Introduction Gas interference in downhole plunger pumps has been studied for severalyears. The first comprehensive analysis was presented by Clegg (1963), whodeveloped a theoretical analysis of separator performance and set some of therules of thumb that are still in use today. These guidelines were applied insubsequent studies that developed practical methods for matching separatorperformance to specific well producing conditions (Campbell and Brimhall 1989;Dottore 1994; Ryan 1992). Poor performance of downhole rod pumps and problemswith progressing cavity (PC) pump operation owing to gas prompted theundertaking of laboratory experimental studies by Robles and Podio (1999) thatincluded visual observation of separator-fluid mechanics using a full-scaleplexiglass wellbore and a conventional rod pump. The problem of downhole gasseparation recently has become of further interest in relation to dewateringlow-pressure gas wells and operating coalbed-methane wells. Patterson andLeonard (2003) studied some different downhole gas-separation designs forcoalbed-methane operations in Wyoming. In these designs, the inlet to the gasseparators was smaller than normal and, along with some baffles, was thought toallow gas to vent from inside the gas separator, obtaining good gas separationin the field installation. While field installations provide the ultimatevalidation of gas-separator performance, it is extremely difficult to isolatethe influence of each design parameter. It was these installations thatprompted the laboratory study of the gas-separator geometry to determinewhether the rules-of-thumb used by the industry for gas-separator design werevalid (Lisigurski 2004). One of the most common sources of inefficiency in oilwell pumpinginstallations (rod pumps and ESPs of PC pumps alike) is gas interference, whichprevents the pump from delivering liquid at the design rate. Although this is awell-known effect, there seems to be limited understanding of the mechanismsthat control gas interference, and this often results in the use of remedies, such as installing downhole gas separators, that are ineffective or evendetrimental to the pumping-system performance. The objectives of this paper are to give a clearer insight on the mechanismsof gas interference in pumping wells and to present the results of recentlaboratory and field studies on the flow characteristics and performance ofsome downhole gas separators. In a pumping installation, one of the principal functions of the wellbore isto operate as a two-phase (gas/liquid) separator so that the pump (which isdesigned to pump liquid) can operate efficiently. Although this concept appearsto be obvious, it seems to be totally ignored by most operators when theydesign completions and install hardware (gas anchors and the like) to combatthe effects of gas interference. In these applications, the separation of gas from liquid is achieved throughgravity separation without the introduction of other mechanisms (centrifugalforces, nozzles, etc.). Thus, the difference in density between the gas andliquid is the main driving force to be used for separation. This also impliesthat forces that oppose the effect of gravity, such as viscous drag caused byhigh fluid velocity and turbulence, will be detrimental to the separationprocess. Thus, high velocity of liquid or gas should be avoided ifpossible.

Laboratory Testing of Downhole Gas Separators

“…Please refer to the literature given in Guzman (2005), Lisigurski (2004), Lisigurski et al (2005), andMcCoy et al (2005).…”