Data processing: generic control systems or specific application – Specific application – apparatus or process – Specific application of pressure responsive control system
Reexamination Certificate
2000-09-11
2004-03-02
Patel, Ramesh (Department: 2121)
Data processing: generic control systems or specific application
Specific application, apparatus or process
Specific application of pressure responsive control system
C700S299000, C700S304000, C700S281000, C700S282000, C700S289000, C700S028000, C700S046000, C700S036000, C702S100000, C702S045000, C702S047000, C137S008000, C137S045000, C137S047000
Reexamination Certificate
active
06701223
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to the optimization of control settings for compressors and/or regulators in a gas pipeline to transition the gas pipeline from an initial state to a target state.
BACKGROUND OF THE INVENTION
Present industry technology for minimizing gas transmission cost in a gas pipeline focuses on optimizing the choice of which of available compressor units to run, and determining values for their pressure control set points. Their objective is to minimize the cost of delivering gas at specified rates and locations while satisfying delivery and supply pressure constraints. The problem and its solution are framed in a steady-state context.
When load or supply conditions change, the optimized solution may no longer apply, and a new set of pressure control set points must be determined. Thus, the operator of the pipeline must transit from the current state of the pipeline to provide for the condition changes. For any given state of the pipeline system and the corresponding load or supply condition changes, the question of realizing an optimum of cost and fuel savings during the transition involves two parts: first, what is a new optimum or target state of the pipeline incorporating the load or supply condition changes, and second, what is the optimum path to attain the new optimum state.
There are software applications available that yield an optimum or target steady state mode of operation for projected load and supply conditions. In addition, there are software applications available for determining the current state of a pipeline system. However, no effective solution has been found for determining the optimum path to reach the optimum state of the pipeline for projected load and supply conditions. Because of significant relaxation times (usually many hours) for long gas pipeline systems, substantial fuel cost will be incurred during the transition to the new optimum state. This cost might be large compared with the anticipated savings at the new optimum state if the transition is not made efficiently.
U.S. Pat. No. 4,835,687 to Martin discloses a system for the optimized management of a pipeline system in real time. In Martin, the control for the pipeline is determined by calculating the pressure and flow rates at important system locations and then determining the corresponding equipment, i.e. compressors, valves, etc., to be used to obtain the calculated pressure and flow rates. The control is then tested to determine whether or not it is feasible for the existing equipment. If feasible, an optimization function is computed taking into account certain optimization criteria such as power consumption and flow rates. The optimization function is a linear combination of various optimization criteria each having corresponding coefficients. The higher the coefficient, the more important the optimization criteria. The optimization function is a function of the state vector that represents the pressures, flow rates and adjustments at important system locations. Convergence of the function is checked and the adjustments to the pipeline system are carried out if the function converges. Optimization of the function occurs using a generalized reduced gradient algorithm. Next, the gradient of the downward processing is checked to see if it is zero. If the gradient is not zero, then optimum points for downward processing of the optimization function are determined and the process is repeated. However, the process is not repeated if the achievement of optimization exceeds the possibilities of real-time computation of the system and optimization is deemed impossible.
If the optimization is deemed impossible due to computational resource constraints, then a pre-computed solution is implemented on the pipeline system, which is not an optimal solution and may not match actual loads.
In practice, Martin suffers several deficiencies. The most serious of these is the inventory problem. Briefly, the inventory problem results from reducing costs over the optimizing transition period. A true optimization will leave the pipeline inoperable at the end of the transition period because inventory (pressure) will have been depleted. It is analogous to the grocer who optimizes profits during the month of March by not replacing inventory sold, so he is out of business April 1.
A second deficiency is that Martin does not teach how to efficiently compute the required gradient. Computation of the gradient from basic principles is impractically slow, requiring thousands of times more computer resources than are available for practical size field problems.
A third deficiency is that Martin does not teach the efficient acceleration of the prescribed generalized gradient iteration. This acceleration combined with an efficient gradient evaluation is essential in solving optimizations in field problems faster than real time.
Therefore, what is needed is a system and technique that addresses these deficiencies and does not require the use of precomputed solutions which may only be available for a very limited number of load combination scenarios.
BRIEF SUMMARY OF THE INVENTION
The present invention determines an optimal set of controls for devices in a pipeline to transit the pipeline from an initial state to a target state over a pre-selected time period, T. A set of controls is a collection of numbers corresponding to a discrete set of time values over interval T for each control station or device in the pipeline system. A control station or device might be a compressor or regulator valve station. The state of the pipeline is the set of pressures and gas velocities at a large representative number of points in the pipeline system. These representative points may be the pressures and velocities at each milepost in the system. (A milepost is simply a marker to define pipeline locations on 1-mile spacing). The target state of the pipeline can be computed in two different ways that can be selected by a user. The target state can be computed externally as a steady state for the pipeline, usually an optimum steady state for the loads (gas deliveries) at the end of time interval T. Alternatively, the target state can be an internally computed state required not to change over a final period &agr;T to T, where &agr; is a suitable number somewhat less than one, but satisfying the prescribed loads at T, and thereby yielding a steady (and thus sustainable) state between &agr;T and T.
The present invention begins by obtaining the initial state of the pipeline from appropriately processed measurements from the field. It generates an initial set of controls and simulates the application of the set of controls on the devices in the pipeline over the pre-selected time period, T, using the initial state information previously obtained and satisfying a set of time-dependent loads (deliveries) at stated points designated throughout the system. The cost for the simulated application of the set of controls is calculated using a cost functional.
A cost functional is a rule for assigning a cost to the set of pipeline states encountered during simulation of the transition from the initial state to time T using a particular control set. For example, at a compressor station the inlet (suction) pressure and discharge velocity and pressure imply a certain theoretical horsepower usage. This theoretical power combined with a machine efficiency and fuel cost imply a cost for the station operation corresponding to this state. Other elements may add to the costs as will be described later. The total cost functional for the simulated period T is the aggregate costs summed over the simulated states achieved during period T.
After the cost for the simulation using a particular control set is calculated, a gradient is determined for that set. The gradient is the vector of numbers comprised of the derivative of the total cost functional with respect to each of the control values in the control set. The set of controls is then modified using the gradient to generate an updated set of controls for the d
Carter Richard G.
Rachford, Jr. Henry R.
Advantica, Inc.
McNees Wallace & Nurick LLC
Pham Thomas
Sattizahn Brian T.
LandOfFree
Method and apparatus for determining optimal control... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for determining optimal control..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for determining optimal control... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3249829