Atlantic Richfield Company

Other

Atlantic Richfield Corporation

Michael D McCormack

Richard F Stoisits

Donald J MacAllister

Tuan D Ma

Kelly D Crawford

Perry W Scherer

Alan D Rock

Chester M Bowling

D Scott Feldman

A computer system and method of operating the same to optimize the operating conditions of a petroleum production field, in which a plurality of wells are arranged according to drill sites, and connected to one or more central processing facilities, is disclosed. In this disclosed embodiment, gas compression capacity is a significant constraint on the operation of the complex production field, and surface line hydraulic effects of well production are to be considered in the optimization. A genetic algorithm is used to generate, and iteratively evaluate solution vectors, which are combinations of field operating parameters such as incremental gas-oil ratio cutoff and formation gas-oil ratio cutoff values. The evaluation includes the operation of an adaptive network to determine production header pressures, followed by modification of well output estimates to account for changes in the production header pressure. Convergence of the genetic algorithm identifies one of the solution vectors as containing an optimal combination of field operating parameters that may be used by production personnel to set the operating conditions of the field.

Petroleum production optimization utilizing adaptive network and genetic algorithm techniques

A system and method is provided for estimating the inverse source wavelet of a seismic shot record for improved accuracy and efficiency in removing high order free surface multiples from the shot record, using a stochastic optimization technique, such as a genetic algorithm procedure, that can more readily accommodate non-linear aspects of this problem. The system and method is applied to a set of measured seismic field data, D.sub.o (x,t), with primary reflections and free surface multiples, where removal of the free surface multiples is determined utilizing an estimate of the inverse of the source wavelet and scale factors made from the seismic data. The system and method automatically optimizes estimates for the inverse of the source wavelet and scale factors.

System for attenuating high order free surface multiples from a seismic

A system for producing a material balance solution for well patterns in a hydrocarbon reservoir is described that automatically optimizes the fluid allocation factors for each well used in determining the solution. The system automatically optimizes estimates for the allocation factors to be used in the material balance solution by randomly generating a first generation of allocation factor strings, each string in the generation assigning allocation factors to each of the wells in the reservoir. A fitness function value is determined for each of the strings by evaluating a fitness function, wherein the fitness function comprises the sum of the differences between computed and measured field pressures for each pattern, and the sum of the differences between target allocation factors and the allocation factors specified within the string for each well. A succeeding generation of allocation factor strings is produced according to a genetic algorithm. The process of determining a fitness function value for each of the strings is then repeated for the succeeding generation. The string having a fitness function value meeting a specified criteria is identified, wherein the identified string represents the optimized estimates of the allocation factors for the reservoir for use in determining the material balance solution.

Automated material balance system for hydrocarbon reservoirs using a

A method of operating an adaptive, or neural, network is disclosed for performing first break analysis for seismic shot records. The adaptive network is first trained according to the generalized delta rule. The disclosed training method includes selection of the seismic trace with the highest error, where the backpropagation is performed according to the error of this worst trace. The learning and momentum factors in the generalized delta rule are adjusted according to the value of the worst error, so that the learning and momentum factors increase as the error decreases. The training method further includes detection of slow convergence regions, and methods for escaping such regions including restoration of previously trimmed dormant links, renormalization of the weighting factor values, and the addition of new layers to the network. The network, after the addition of a new layer, includes links between nodes which skip the hidden layer. The error value used in the backpropagation is reduced from that actually calculated, by adjusting the desired output value, in order to reduce the growth of the weighting factors. After the training of the network, data corresponding to an average of the graphical display of a portion of the shot record, including multiple traces over a period of time, is provided to the network. The time of interest of the data is incremented until such time as the network indicates that the time of interest equals the first break time. The analysis may be repeated for all of the traces in the shot record.

Adaptive network for automated first break picking of seismic refraction

A genetic algorithm method of scheduling the delivery of coal to a series of incoming coal trains is disclosed, where each of the coal trains corresponds to a coal recipient having different premium and penalty rates for energy yield, and possibly also different contaminant specifications. The method is applied to a coal facility having a plurality of bins, where the coal is of varying quality at the various bins, and where the bins are grouped so that loading into trains must be done in succession. According to the disclosed method, a fitness function is generated, with adjustable weighting factors, to combine the premium or penalty revenue with any latency or delay time, as well as moves of the coal loading plow for the series of slots. A randomly selected first generation of assignment sequences are evaluated according to the fitness function, with succeeding generations produced according to genetic techniques, until convergence to the optimum assignment sequence is reached. The method may also be applied to other situations in which non-uniform inventory is to be delivered to recipients having different requirements.

Genetic method of scheduling the delivery of non-uniform inventory

An adaptive, or neural, network and a method of operating the same is disclosed which is particularly adapted for performing seismic trace editing for seismic shot records. The adaptive network is first trained according to the generalized delta rule. The disclosed training method includes backpropagation is performed according to the worst case error trace, including adjustment of the learning and momentum factors to increase as the worst case error decreases. Slow convergence regions are detected, and methods applied to escape such regions including restoration of previously trimmed dormant links, renormalization of the weighting factor values, and the addition of new network layers with links between nodes that skip the hidden layer. After the training of the network, data corresponding to a Discrete Fast Fourier Transform of each trace, and to certain other attributes of the trace and adjacent traces thereto, are presented to the network. The network classifies the trace as good or noisy according to the inputs thereto, and to the weighting factors therewithin, such classification useful for ignoring noisy traces in subsequent data analysis. The analysis may be repeated for all of the traces in the shot record, and in multiple shot records.

Method for the automated editing of seismic traces using an adaptive

An adaptive, or neural, network and a method of operating the same is disclosed which is particularly adapted for performing first break analysis for seismic shot records. The adaptive network is first trained according to the generalized delta rule. The disclosed training method includes selection of the seismic trace with the highest error, where the backpropagation is performed according to the error of this worst trace. The learning and momentum factors in the generalized delta rule are adjusted according to the value of the worst error, so that the learning and momentum factors increase as the error decreases. The training method further includes detection of slow convergence regions, and methods for escaping such regions including restoration of previously trimmed dormant links, renormalization of the weighting factor values, and the addition of new layers to the network. The network, after the addition of a new layer, includes links between nodes which skip the hidden layer. The error value used in the backpropagation is reduced from that actually calculated, by adjusting the desired output value, in order to reduce the growth of the weighting factors. After the training of the network, data corresponding to an average of the graphical display of a portion of the shot record, including multiple traces over a period of time, is provided to the network. The time of interest of the data is incremented until such time as the network indicates that the time of interest equals the first break time. The analysis may be repeated for all of the traces in the shot record.

Individual seismic source units of a surface array activated simultaneously for cumulative reinforcement may initiate seismic waves at slightly different times despite efforts at synchronization. These variations are measured and used to compute an inverse operator which is applied to the summed uncorrected seismic recording so as to undo the distortion they produce.

Growing community of inventors

Plano, TX, United States of America

Michael D McCormack