Practical control of timestep selection in thermal simulation

Conference · Tue Nov 01 04:00:00 EST 1983 · Soc. Pet. Eng. AIME, Pap.; (United States)

OSTI ID:5264978

Sammon, P H; Rubin, B

This paper describes an error driven timestep selection scheme that utilizes a new technique for estimating time truncation error. The calculation of this estimate uses only current reservoir information. Thus, information from past timesteps, such as that required by the differencing technique of Mehra et al or Jensen, is not required. This consideration is particularly important if past reservoir states do not correspond well to the current one, such as might occur when relatively large timesteps are being taken or when a model formulation change occurs. Once the time truncation error estimate is obtained, techniques from Lindberg are used to calculate an appropriate timestep. The calculation introduces little additional simulator overhead and can be easily integrated into any fully implicit reservoir simulator. An implementation of the above timestep selection technique is described for a thermal reservoir simulator. A provision for limiting the timestep by user input desired changes is provided and a method for estimating the results of certain Newton iterations using timestep information is described. The timestep selection scheme also has the ability to calculate a timestep at well changes, relieving the user from the necessity of estimating these parameters. The timestep selection algorithm is intended to afford greater simulator accuracy and reliability. The algorithm's robustness also affords greater efficiency when compared to damped schemes such as that presented in Grabowski et al. The parameters controlling the error-driven timestep calculation are easily selected by the user. It is typically found that the new method can be quickly made to perform as well or better than a simulation run using desired changes with carefully selected parameters. Also, the accuracy of the results obtained with the new method are superior.

• Research Organization: Computer Modelling Group
• OSTI ID: 5264978
• Report Number(s): CONF-831190-
• Conference Information: Journal Name: Soc. Pet. Eng. AIME, Pap.; (United States) Journal Volume: SPE11268
• Country of Publication: United States
• Language: English

Related Subjects

02 PETROLEUM
020300* -- Petroleum-- Drilling & Production
ACCURACY
ANALOG SYSTEMS
CALCULATION METHODS
ENHANCED RECOVERY
FUNCTIONAL MODELS
MATHEMATICAL MODELS
RECOVERY
RESERVOIR ROCK
SIMULATION
SIMULATORS
THERMAL RECOVERY