Welcome Guest - Navigation: Signup | Login | Home

GASDECSYS: A Decision Support System for Gas Distribution Networks

GASDECSYS simulates flows and pressures in a gas distribution networks; it computes least cost reinforcements to meet increasing demands; and it proposes supply policies to minimize purchase costs.

The modeling of the design and operations of gas distribution networks raise many technical difficulties due to the non-linearity of the physical transmission laws. Thanks to a new and innovative approach, GASDECSYS can handle large scale real-life problems online, which makes it a versatile tool to explore on-line a variety of situations. The prototype version proposed here is based on the open-source solvers of the cvxopt library. For increased stability and the handling of very large scale instances, the fully fledged version will use an in-house solver.

 
Login or Sign up for free

and
 

» To learn more about GASDECSYS, explore the prototype version »
GASDECSYS can be tailored to your needs. Please contact us.

 

To experiment with GASDECSYS, the present web interface proposes a demonstration on illustrative examples, with the flexibility of changing components of the networks. The tool, based on the prototype version, is made of three interactive modules

  1. The Simulator module provides target values for the pressures at the inlets and outlets of the network, as well as compressors and regulators operating ratios, to meet supply and demand requirements.
  2. The Reinforcement module proposes investments in pipe reinforcement when the existing network cannot meet the supply and demand projections.
  3. The Supply Management module suggests efficient purchase and selling strategies that are compatible with the existing network.

GASDECSYS computes flows and pressures according to the nonlinear law of stationary flows of compressible fluids.

If you wish to know more about

Specifics of the demo version

  • GASDECSYS is based on the scientific contribution [4].
  • GASDECSYS computes the stationary flows and the related pressures in a pipe of length L and inner diameter D according to Weymouth law [8], or more accurately to its approximation
    LβQ2 --5 = (pi)2 - (po)2,D
    where Q is the flow through the pipe, pi and po are the pressures at the inlet and the outlet of the pipe, respectively, and β is a technical coefficient. In the approximation β is taken to be independent of the pressures and the flow. This is considered to be acceptable for design purposes. The numerical value for β is taken from [6].

    The compressors are characterized by their compression ratio, which is the ratio between the discharge pressure and the suction pressure.

  • The purchase and selling costs are inspired from [6].
  • The investment cost function is of the form
    2 C (D,L ) = (k2D  + k1D  + k0)L.
    The value of the coefficients k2, k1 and k0 are to be found in [6].
  • The network examples are inspired from published data. The characteristics of the network (pipe lengths, compressors and regulators locations, bounds on the flows at the inlets and outlets, bounds on the pressures) come from the cited references. The coefficient β and the cost function C(D,L) are those described above. The “simple network” appears in [1] and studied in [7]. The “Belgium network” appears in [56]. The “larger network” appears in [7] and is inspired by a real network at Gaz de France.
  • Limitations of the demo version:
    • Maximum number of nodes: 30
    • Maximum number of arcs (including compressors and regulators): 50
    • Maximum pipe diameters: 1200 mm
    • The demo version uses an open-source solver with a maximum number of iterations of 200.

About the Simulator module

The Simulator module provides target values for the pressures at the inlets and outlets of the network, as well as compressors and regulators operating ratios, to meet supply and demand requirements.

Compressors and regulators in the network create options in the operating mode. Each one of them call for power supply to activate the compressors within the network and/or at the terminals. A commonly used expression [12] for the power supply cost of a compressor is of the form

          (            )
            (pd )γ2
W   = γ1Q    ---   -  1  ,
             ps
where ps and pd are the suction and discharge pressure, respectively; and γ1 and γ2 are parameters with γ2 1.18. The demo version does not use this formula but a surrogate for it.

Compressors impose a flow orientation. However, depending on the actual supply and demand, it may be necessary to change this orientation. The demo version searches for the best orientation to meet the current supply and demand needs.

About the Reinforcement module

The reinforcement module proposes a network reinforcement to meet increased demands. It balances the investment cost (installation of new pipes) and the power supply cost required to enforce the necessary flows. While investment costs are unambiguously known, the power supply operations costs are much uncertain, due to the seasonal and stochastic variations of demands. We believe that expert judgment is required to compare several alternatives proposed by the module. To this end, the user can vary a parameter, named weight on the reinforcement cost, to control the intensity of the investment. For each investment proposal, the simulation module computes the full flow solution and compressor requirements.

About the Supply Management module

The supply management module suggests efficient purchase and selling strategies that are compatible with the existing network. It balances the supply cost and the power supply cost required to enforce the necessary flows. To this end, the user can vary a parameter, named weight on supply cost, to control the intensity put on the supply cost.

Purchase and selling strategies cannot be elaborated without concern to the seasonal and trend variations of the demand; storage facilities allow smoothing but call for a dynamic representation of the problem. The demo version does not take into account the time structure of the problem, but it is an essential building block in the tool for the design of a purchase policy.

About the fully fledged version

Offline features of GASDECSYS and developments in progress:
  • General design module (e.g. design from scratch).
  • Installed diameters restricted to commercial sizes.
  • Incorporate power supply costs such as those in [12].
  • Incorporate seasonal effects to plan a dynamic supply/delivery strategy.
  • Replacement of the open source solver by a dedicated in-house one.

References

[1]   M. Abbaspour, K. Chapman, and P. Krishnaswami. Nonisothermal compressor station optimization. Journal of Energy Resources and Technology, 127(2):131–141, 2005.

[2]   J. André and J. Bonnans. Optimal features of gas transmission networks. In J. Herskovits, editor, Proc. Eng. Opt. 2008. International Conference on Engineering Optimization, Rio de Janeiro, June 1-5, 2008, 2008.

[3]   J. André, F. Bonnans, and L. Cornibert. Planning reinforcement on gas transportation networks with optimization methods. European Journal of Operational Research, 2008 (forthcoming).

[4]   F. Babonneau, Y. Nesterov and J.P. Vial. Design and operations of gas transmission networks. Operations Research, to appear.

[5]   D. de Wolf and Y. Smeers. Optimal dimensioning of pipe networks with application to gas transmission networks. Operations Research, 4(44):596–608, 1996.

[6]   D. de Wolf and Y. Smeers. The gas transmission problem solved by an extension of the simplex algortihm. Management Science, 46(11):1454–1465, 2000.

[7]   F. Tabkhi. Optimisation de réseaux de transport de gaz. PhD thesis, Institut National Polytechnique de Toulouse, France, 2007.

[8]   T. R. Weymouth. Problem in natural gas engineering. ASME Transactions. 34:185–234, 1942.