A pressure pulse, known as a water hammer, can occur immediately after water injection wells are shut-in for emergency or operational reasons. Large pressure pulses may cause wellbore integrity problems such as sand-face failure and sand production. We propose a new workflow to simulate water hammer events, the resulting wellbore failure and sand production in water injectors. Based on the results of this workflow, recommendations are made for wellbore design and shut-in protocols for water injection wells. The results presented in this paper, for the first time, allow us to quantitatively understand the role of well shut-downs and subsequent water hammer pressures on sand production. The failure of unconsolidated sands near the wellbore is affected by water hammer events, their amplitude, period, and attenuation. If a water hammer event occurs during shut-in of water injectors, the extent of the sand failure becomes larger and the failure zone continues to propagate along the stress concentration direction. The simulation results clearly show which parameters are important and suggest changes to well operations such as proper shut-in protocols that help to minimize the possibility of sand production. The results also suggest ways in which injectors can be designed to minimize the impact of water hammer events.

Abstract Water hammer effects resulting from the shutting in of water injection wells are an often ignored issue in petroleum production operations but they have considerable impact on injection well performance and longevity. Mismanaged, they can result in substantial and perhaps irreparable damage. This paper presents a study on the creation and propagation of water hammer due to rapid shut-in of water injectors. Water hammer 1–4 or pressure surge, is a pressure transient phenomenon which has long been known to occur as a result of a sudden change in fluid flow velocity. In water injectors, rapid shut-in creates a water hammer. Over time, injectors that undergo repeated rapid shut-ins often have significantly reduced injectivity and show evidence of sanding and even failure of the down-hole completion 5 . It is therefore critical to understand the nature of water hammer including the magnitude, frequency, and energy dissipation. To study the water hammer in water injectors, a field trial was conducted to record pressure pulses generated from rapid shut-ins, at different well depths, in a soft formation, cased and perforated (C&P) water injector. Modeling work was conducted to understand the data. The results of the field trial and model work demonstrated that: The magnitude of the first pressure pulse due to abrupt shut-in can be estimated by using the equation: ?p = V×??×c, where V is the flow velocity, ??r is the fluid density, and c is the speed of propagation of a pressure signal along the well. High rate sampling (up to 100 samples/second) is required to capture subtle details of the water hammer signal. Water hammer can be modeled as a low-frequency pressure wave similar to the higher frequency Stoneley waves produced during VSP and by acoustic logging tools. Models of water hammer propagation in a synthetic analog of the test well reproduced most of the details of the signals recorded during the tests. Introduction Water injectors have become increasingly important in recent years as maturing oil production has come to rely on water injection to maintain high rates from reservoirs with low reservoir energy and sand production issues. There are many reported studies of sand production in producers 6–12 but very few discussions of sanding issues in water injectors 5, 13–15 . Several factors, including repeated cycles of injection and shut-down, crossflow, backflow, and water-induced strength reduction, are known to cause sanding in injectors. One important but under-studied factor is the damage induced by water hammer (WH) pressure pulses, which may produce cyclic pressure variations of hundreds of psi. In weak sands, fluid pressure fluctuations in the near-wellbore as small as a few tens of psi may be sufficient to cause sand failure 14 . A recent modeling study 5 , summarized in Figure 1, shows that for an unconsolidated rock the combination of WH and injection pressure creates more sanding than a monotonic increase in injection pressure with the same equivalent total injection pressure. The additional sanding is attributed to the faster rate of formation degradation in response to the more frequent pressure pulses, which is demonstrated in Figure 2.