Since many gas wells in the North Sea are at the end of production life, liquid loading is a serious issue that the industry is facing. The gain in gas production that would result from preventing or delaying liquid accumulation is significant [Kreft, 2009]. TNO has a long term track record in collaboration with different gas operators on finding and evaluating solutions to this problem [Schiferli, 2010; Alberts, 2012].
One of the promising mitigation methods is the use of foamers to deliquify gas wells. The surfactants transform the liquids into foam, which is much easier to transport to the surface (due to the reduced density) and therefore postpone the occurrence of liquid loading, see Figure 1. Drawbacks of the use of foam, however, are the difficulties in the downstream handling of the foam and the possible environmental impacts of foam (which, amongst others, put limitations to the maximum amount of foamer that can be used). According to the industry, foamers, which do foam in the current laboratory tests, may not always perform well in the field. Besides the difficulties in the selection of the foamers, the efficiency of foamers is also challenging, since the density reduction is uncertain. This poses problems for the operators, since the mitigation method becomes less predictable.
Figure 1 : Left graph: snapshot of a movie showing a gas well that is being deliquified using surfactants [Miller 2009]. Right graph: pressure gradient tubing performance curve with various surfactant concentrations [Nimwegen 2015]. For a foamer concentration of 1000ppm, liquid loading is postponed to about 6m/s and the pressure gradient is reduced by about 60%.
The state-of-the-art knowledge on using foamers for the deliquification of gas wells is mainly based on empirical data [e.g., Bell, 2008; Solesa, 2006; Willis, 2008]. An average reduction of the critical velocity by the application of a surfactant is about 50% with respect to the unfoamed condition, according to [Jelinek 2005]. This value is based on field experience and taken as a rule of thumb in their foamer applications. A better prediction method could be the criterion of Richter (1981) (). In order to qualify a foamer for application in the field, both the gas operator as well as the foam suppliers use test methods that are derived from a selected number of available test methods (ASTM D-3519, ASTM D-892, ASTM D-3601). These test methods are all conducted at atmospheric pressure and at lower temperatures than encountered in gas wells. Only a small number of researchers have attempted to look at foam performance at higher pressures and temperatures. As a consequence, foamers, which do foam in the current laboratory tests may not always perform well in the field. The uncertainty is mainly due to the large difference in conditions used in the lab and the field. Firstly, if a field trial fails, the fact that testing cannot be performed at actual field conditions makes it difficult to evaluate the cause of failure in the field [Veeken et al. 2012]. Secondly, foamers may be unfairly excluded on the basis of their performance at ambient conditions. Inclusion of these effects requires laboratory tests at downhole conditions, which are currently not available. Finally, a correlation between the results of these test methods and the effect on, e.g., the critical velocity is not known.
In literature some knowledge on the rheology of foam [Weaire, 2008; Verbist 1997; Cates 1997] is available. More specific knowledge about the parameters that influence the foam stability performance [Maini, 1986] is scarce. Especially, the knowledge on the importance of the key parameters pressure and temperature, that potentially influence the foam performance for deliquification applications, is very limited.
To improve the understanding of foam for deliquification purposes TNO started in 2013 the TKI-Upstream Gas project: Foam I “Experimental Foam Evaluation, supported by NAM, Total, GdF (now Engie), TAQA, ONE and EBN.
Foam I started with interviewing the operators and suppliers on their experience in the use of foamers for deliquification. Based on this outcome a desktop scale (Bikerman) sparging column has been designed to test the ‘foamability’ of a range of solutions, see Figure 2. In this setup N2 is sparged through a premixed solution thereby generating foam. The foam build-up (foam formation rate), foam collapse (after stopping the sparging), and liquid volume overflow (or carry-over) has been measured in these tests. The setup is capable of testing up to 17barg and 150°C, hence approaching field conditions. The premixed solutions included a variation of : surfactant concentration (500-8000 wppm), salinity (NaCl : CaCl2 = 4:1), hydrocarbon fraction (C12, 0-90 wt%). CO2 was dissolved in all solutions to obtain pH = 4. For these tests, only Nitrogen was used as indicated by literature that the effect of sparged gas on the ‘foamability’ will mainly affect the foam coarsening [Saint-Jalmes et al. 2002]. The tests were performed at various temperatures, pressures and sparging flow rates. The highlight of the results obtained from the desktop-scale tests was the marginal effect of pressure on the foamability and drastic effect of temperature on the foam formation and stability. The outcome of the desktop scale experiments have been put into a new proposal for foamer selection guidelines and have been presented in a workshop to operators and chemical vendors.
Figure 2 : Left graph: Desktop scale (Bikerman) sparging column. Right graph: mid-scale flow loop (TU Delft).
Simultaneously, flow tests have been performed in the 50mm mid-scale flow loop located at the (former) “Kramers Laboratorium voor Fysische Technologie” of the Delft University of Technology, see figure 2. This setup has been designed by Dries van Nimwegen for his PhD research. The flow loop consist of a 12m vertical acrylic pipe of 50mm ID, built out of sections with varying length (in the range of 0.3m to 1m length). All pipe sections are interconnected using flanges and are machined to ensure smooth transitions. Dry air is flowing into the setup at the bottom, and is released to atmosphere at the top of the setup. Liquid is injected into the system downstream of the gas inlet, dragged along with the gas flow and is collected at the top of the setup to be reinjected in the system. The injected liquid is a premixed water-surfactant-solution and foam is generated by the flow. The flow loop is equipped with pressure sensors to measure the pressure gradient and two fast-closing valves to measure the liquid holdup and mean foam density. The tests were performed at atmospheric conditions, with various surfactant concentrations (0 – 2000 wppm) and various gas flow rates. The mid-scale flow loop results obtained in this project, together with the data obtained by van Nimwegen (2015), have been used to develop a preliminary predictive model of flow, which has been put in a GUI so that it can be used in the workflow of operators in their flow assurance (see figure 3)
Figure 3 : GUI developed in Foam I that is able to compute the effect of foamers on a well. Realistic well parameters can be chosen, and the calculation results can be exported to a PROSPER readable format. The computation of a single TPC is done in several seconds.
The industry, both operators and suppliers, have recognised Foam I to have made good steps forward in development of a foam selection guideline and a predictive flow model. However, some of the testing requirements as been outlined in the foam selection guidelines by Foam I (i.e. testing at high temperature) are regarded as costly to be included in the current protocols (discussed during the workshop with operators and chemical vendors). The industry have also suggested that a validation of the foam selection guidelines with ‘real’ hydrocarbons and ‘real’ brines should be performed. Finally, the current predictive flow model uses input data derived from the mid-scale tests, which are costly, time-consuming and difficult to obtain in combination with real-field fluids. Foam II aims to address the above items by considering the fundamental mechanisms of the foamer action in more details, ‘typical’ chemical compositions of the fluids encountered in production, and validation with field data.
 The Richter criterion predicts for the 50mm lab tests an onset of liquid loading at 13m/s for an unfoamed condition and 7.5m/s at 1000ppm foamer concentration; the lab results are 15m/s and 6m/s, respectively. Using this criterion for a typical well configuration (BHP = ~35bar, rg = ~30 kg/m3, rl = ~1000 kg/m3, s = ~0.06 N/m,
D = ~0.15), the Richter criterion predicts an onset of liquid loading at 147kNm3/d for an unfoamed condition and 95 kNm3/d at 1000ppm foamer concentration. Although the Richter criterion seems adequate and readily available, its most significant short-coming is that it needs an estimate of the reduction of the liquid density due to the addition of foamer.