The Irreversible Port-Hamiltonian Systems (IPHS) framework is extended to the modelling of non-isentropic fluids with viscous dissipation in the Eulerian description. Building on earlier IPHS formulations for diffusion-driven and non-convective distributed systems, it is shown that convective transport can be consistently encompassed by the framework by modifying the underlying differential operators. After revisiting the constitutive relations of non-isentropic fluids in both Eulerian and Lagrangian coordinates, it is demonstrate how these systems fit within an extended IPHS formulation. Furthermore, an extended parametrisation of the boundary port variables which ensures that the first and second laws of Thermodynamics are fulfilled allows to define a general class of boundary controlled IPHS.
Introduction
Irreversible port-Hamiltonian Systems (IPHS) formulations were first introduced in [11 b] and later developed as an alternative to pseudo port Hamiltonian formulations [3, 5] or metriplectic-GENERIC [4] formulations, to extend classical port-Hamiltonian theory to the modelling and control of irreversible thermodynamic systems whose dynamics depend strongly on temperature. Such systems include many cutting-edge engineering applications involving active materials, fluid flows, or energy conversion processes (e.g., chemical or nuclear reactions) [8, 2]. In contrast to classical port-Hamiltonian approaches, IPHS use the total energy as the potential function and incorporate the entropy balance directly into the system dynamics. This leads to a nonlinear structure that depends on the co-state variables and no longer satisfies the Jacobi identities. Nevertheless, the modulated skew-symmetric interconnection matrix still reflects energy conservation in accordance with the first law of thermodynamics, while the entropy balance enforces the second law. A nice property is that the modulation function explicitly depends on the topology and the irreversible thermodynamic forces of the irreversible processes involved in the systems. These features make IPHS a very promising framework, particularly for control design [13], which can be reinterpreted in terms of energy shaping and entropy assignment. IPHS formulations have recently been extended to distributed-parameter systems defined over one-dimensional spatial domains [9, 10]. This extension has proven particularly effective for representing diffusion-type systems, such as the heat equation, and for deriving boundary control laws that ensure their stabilization [6]. The same parametrization can be applied directly to non-convective hyperbolic systems, such as deformable mechanical structures, but it is not suitable for convective systems. In the latter case, the assumption of local thermodynamic equilibrium requires tracking the material particles, which naturally leads to formulations in the Lagrangian framework. The aim of this paper is to extend the IPHS formulations to convective systems formulated in the Eulerian description, focusing on the simple yet illustrative case of a non isentropic fluid with viscous damping.
The paper is organized as follows: § 1.1 provides the governing equations of 1D non isentropic compressible fluids in both Eulerian and Lagrangian coordinates. In § 2 are derived the irreversible port Hamiltonian formulations in these two different frames, leading to the general IPHS formulation in § 3. Some conclusions and perspectives are given in Section 5.
Non-isentropic compressible fluids
In this section, we focus on non-isentropic compressible fluids within a one-dimensional spatial domain. The behavior of fluid flows can be described from two different reference frames. In the Eulerian frame, or fixed coordinate system, the focus is on observing the fluid as it moves past a stationary point in space. In contrast, the Lagrangian frame, or moving coordinate system, follows the motion of individual fluid particles. Figure 1 vividly depicts the differences between the two coordinate frames. Now, we derive the governing equations of fluid flow using mass, momentum and energy (entropy) balance in both the reference frames.
Figure 1. Eulerian and Lagrangian coordinates.
Governing equations : Eulerian coordinate
We denote by ρ=ρ(ζ,t), v=v(ζ,t), p=p(ζ,t) the fluid mass density per unit length, velocity and pressure, respectively, with ζ∈[a,b]. By applying Reynold’s transport theorem and divergence theorem, the conservation of mass can be represented as the continuity equation2
∂t∂ρ+∂ζ(ρv)=0.(1)
Next, using Newton’s second law for the material volume neglecting the gravitational forces and then applying Reynold’s transport theorem and divergence theorem, we obtain the Cauchy momentum equation
∂t(ρv)+∂ζ(ρv2)=∂ζσ,(2)
where σ represents the stress tensor. Consider σ=−p−τ, where p is the thermodynamic pressure, τ is the viscous stress. Now, considering a Newtonian fluid, we have the viscous stress τ=−μ∂ζv, where μ is the coefficient of dynamic viscosity. Using the continuity equation (1) we have the following momentum balance equation
ρ∂tv=−ρv∂ζv−∂ζp+∂ζ(μ∂ζv).(3)
Now, using Gibbs’ equation, i.e. du=−pdΦ+Tds where Φ=ρ1 is the specific volume, we have
∂ζu=−p∂ζ(ρ1)+T∂ζs.(4)
By using the specific enthalpy h:=u+ρp we have
ρ1∂ζp=∂ζh−T∂ζs.(5)
From (3), we have
∂tv=−v∂ζv−∂ζh+T∂ζs+ρ1∂ζ(μ∂ζv).(6)
Finally, the rate of change of the internal energy per unit mass u is equal to the sum of the heat conduction due to the heat flux q and the mechanical work, and is given by
∂tu=−v∂ζu−ρ1∂ζq−ρp∂ζv−ρτ∂ζv,(7)
The heat flux follows Fourier’s law, given by q:=−k∂ζT where k is the thermal conductivity. Using σ=−p+τ, we have
∂tu=−v∂ζu+ρ1∂ζ(k∂ζT)−ρp∂ζv+ρμ(∂ζv)2.(8)
Hence, (1), (6), and (8) constitute the governing differential equations of a compressible non-isentropic fluid in Eulerian coordinates flowing in one dimensional spatial domain. Again, using Gibbs’ equation we have3
Dts=T1Dtu+TpDt(ρ1).(9)
where Dt:=∂t+v∂ζ is the material derivative. Then, expanding the material derivative, we have the following governing equation in terms of specific entropy s instead of specific internal energy u
∂ts=−v∂ζs+ρT1∂ζ(k∂ζT)+ρTμ(∂ζv)2.(10)
To summarize, the governing equations of a non isentropic fluid subject to viscous damping and heat conduction in Eulerian coordinates are:
∂tρ=−∂ζ(ρv)(11)
∂tv=−v∂ζv−∂ζh+T∂ζs+ρ1∂ζ(μ∂ζv)(12)
∂ts=−v∂ζs+ρT1∂ζ(k∂ζT)+ρTμ(∂ζv)2.(13)
and the total energy is given by
H(ρ,v,s)=∫ab(21ρv2+ρu(ρ,s))dζ
Governing equations: Lagrangian coordinate
Consider now a small fluid element as in Figure 2.
Figure 2. Small fluid element in spatial and material coordinates.
By performing similar calculations as outlined above and applying the definition of the material derivative, we can derive the velocity and entropy equations in Lagrangian coordinates as follows:
Dtv=−ρ01∂ηp+ρ02μ∂η2v(18)
Dts=ρ0T1(k∂η2T+μ(∂ηv)2).(19)
Then, (17), (18), and (19) constitute the governing equations of a compressible non-isentropic fluid in Lagrangian coordinates flowing in one dimensional spatial domain via replacing the material derivative Dt with ∂t.
IPHS FORMULATIONS OF NON ISENTROPIC COMPRESSIBLE FLUIDS
In this section we derive the IPHS formulation of non isentropic fluids in both Eulerian and Lagrangian coordinates.
IPHS formulation in Eulerian coordinates
In IPHS formulations [9], the transport and dissipative mechanisms are represented in a way that explicitly highlights energy flows and entropy production. This explicit representation is crucial for both physical interpretation and control design. With this goal in mind, we choose as state variables the energy variables x:=(xTs)T, where x:=(ρv)T. The total energy (Hamiltonian) of the fluid is defined as
H(x)=∫ab(21ρv2+ρu(ρ,s))dζ(20)
=∫ab(2ρv2+ρh−p)dζ.(21)
The co-energy (effort) variables are defined as
δρH=21v2+h
δvH=ρv
δsH=δuδH∂s∂u=ρT(since du=−pd(ρ1)+Tds)
Now, we can rewrite the governing equations (1), (6), and (10) in terms of energy and co-energy variables as follows
encoding the second law of Thermodynamics, i.e. dtdS≥0 if the boundary conditions are set to zero.
We will see in Section 4 that the system (28) with appropriate definition of the input/output map fits with the generalisation of the IPHS formulations proposed in [9] to systems with convective flows.
IPHS formulation in Lagrangian coordinates
Consider now the same compressible, non-isentropic fluid in Lagrangian coordinates whose dynamics is governed by the equations (17), (18), and (19). We choose as state variables z:=[zTs]T, where z:=[ϕ:=1/ρv]T. The Hamiltonian is given by:
H(z)=∫ab(21ρv2+ρu)dξ=∫ab(21v2+u)ρ0dη.
Then the co-energy variables are:
δϕH=−ρ0p,δvH=ρ0v,δsH=ρ0T.
It can be easily verified that in Lagrangian coordinates, the viscous stress reads τ=−ϕρ0μ∂ηv, and the heat flux q=−ϕρ0k∂ηT.
We can now rewrite the governing equations in terms of energy and co-energy variables in Lagrangian coordinates as
Where the operators A1 and B1 are defined from Hilbert space H1([a,b]) to H1([a,b]) as follows:
A1:f↦ρ01∂η(ϕTρ02μ∂ηvf)(32)
B1:g↦Tϕρ02μ∂ηv∂η(ρ01g)(33)
and the operator C1 is given by
C1δsH=ρ0T1∂η(ϕρ0k∂η(ρ01δsH)).(34)
Lemma 3. The operator B1 is the formal adjoint of A1, i.e. B1=A1∗. Moreover, C1 is formally skew-symmetric.
Proof. It is easy to show that Tϕρ02μ∂ηv∂η(ρ01⋅) is the formal adjoint of ρ01∂η(Tϕρ02μ∂ηv⋅) by using integration by parts considering homogeneous boundary conditions. The skew adjointness of ρ0T1∂η(ϕρ0k∂η(ρ01⋅)) can be proven in a similar way. □
encoding the second law of Thermodynamics, i.e. dtdS≥0 if the boundary conditions are set to zero.
In this case the system (35) with appropriate definition of the input/output boundary maps fits with the formulations given in [9] and their generalisation to convective systems proposed in Section 4.
Boundary controlled IPHS
Building on the previous examples, the IPHS definition for boundary-controlled systems in [9] can be extended to include irreversible phenomena and convective flows.
Definition 4.1. A general structure of an infinite-dimensional irreversible port-Hamiltonian system with potential convective flows undergoing m irreversible processes is defined by
where ∂ζf(g)=f(x)∂ζ(f(x)g), R0,R1∈Rm×1, rs∈R are the modulating functions defined as follows
R0,i=γ0,i(x,δxH,δsH){S∣G0(:,i)∣H},
R1,i=γ1,i(x,δxH,δsH){S∣G1(:,i)∣∂ζfH},
rs=γs(x,δxH,δsH){S∣H},
γ0,i,γ1,i,γs are strictly positive-valued functions. G0,G1∈Rn−1×m, C(x)∈Rm×1. The pseudo-Poisson brackets are defined as
{E∣G∣F}=[δxETδsET][0−G∗G0][δxFδsF],
{E∣gs(x)∣F}=δsE∂ζ(gs(x)δsF)
for some smooth function E and F, operator G and its formal adjoint G∗ and a real continuous scalar function gs(x).
From this definition one can parametrize sets of boundary input and output in order to define the considered class of boundary controlled IPHS.
Assumption 1. The operator
[δxHδsH]T[P1∂ζf(⋅)−G1TC(x)]δxH
is an exact differential form.
Definition 4.2. A Boundary Controlled IPHS (BC-IPHS) is an infinite dimensional IPHS according to Definition 4.1 augmented with the boundary port variables
with R(x,δxδH)=[1R1(x,δxδH)rs(x,δxδH)] and
WB=[21(Ξ2+Ξ1Pep)Mp21(Ξ2−Ξ1Pep)Mp],
WC=[21(Ξ1+Ξ2Pep)Mp21(Ξ1−Ξ2Pep)Mp],
where Mp=(MTM)−1MT, Pep=MTPeM and M∈R(n+m+2)×k is spanning the columns of Pe∈Rn+m+2 of rank k, defined by 4
Pe=P10G1T0000gsG10000gs00(41)
and where Ξ1 and Ξ2 in Rk×k satisfy Ξ2TΞ1+Ξ1TΞ2=0 and Ξ2TΞ2+Ξ1TΞ1=I.
A direct consequence of the structure given in Definition 4.1 and 4.2 is the completion with the first and second law of Thermodynamics as stated in the following lemmas.
First law of Thermodynamics
Lemma 4 (First law of Thermodynamics). The total energy balance is
dtdH(t)=y(t)Tv(t)(42)
which leads, when the input is set to zero, to ∂H=0 in accordance with the first law of Thermodynamics.
Second law of Thermodynamics
Lemma 5 (Second law of Thermodynamics). The total entropy balance is given by
dtdS(t)=∫abσtdξ+ys(t)Tvs(t)(43)
where ys and vs are the entropy conjugated input/output and σt is the total internal entropy production. This leads, when the input is set to zero, to dtdS=∫abσtdξ≥0 in accordance with the second law of Thermodynamics.
Proof. The proof follows the same development as in [9] with the additional use of Assumption 1 which accounts for the coupling between the material and thermal domain due to the convection.
The formulation of non-isentropic, non-reactive fluid defined in Eulerian and Lagrangian coordinates is given in the following propositions.
Proposition 4.3.A compressible, non-isentropic fluid in Eulerian coordinates can be described as an irreversible port-Hamiltonian system (38) considering the following boundary inputs and outputs
where fs:=ρk∂xT is the entropy flux, and the irreversible entropy production rate is
S˙=∫0Lσsdξ+ysTus,
where S is the total entropy of the fluid, us and ys are the boundary entropy input and output respectively. Moreover, σs≥0 represents the total internal entropy production.
Proof. For a non-isentropic, non-reactive fluid defined in Eulerian coordinates, we have x=[ρv]T, S=∫0bρsdξ, δxS=[s0ρ]T. The matrices in (38) are P0=02×2, P1=[0−1−10], G0=02×1, G1=[01]T, C(x)=∂ζs, gs=1. We will have:
We can compute R1=Tμ∂τv, R0=0, rs=T2k∂τT, and the irreversible entropy production rate is γ1=Tμ>0, γ0=0, γs=T2k>0. We have f0=1 and f=ρ1 leading to the extended set of effort variables
where σs:=Tμ(∂ξρδvH)2+T2k(∂ξρδsH)2≥0 is the internal entropy production, ysTus:=(spv+Tq)ab. □
Proposition 4.4. A compressible, non-isentropic fluid in Lagrangian coordinates can be described as an irreversible port-Hamiltonian system (38) considering the following boundary inputs and outputs
where us and ys are boundary entropy input and output respectively. Moreover, σs≥0 represents the total internal entropy production.
Proof. For a non-isentropic, non-reactive fluid defined in Lagrangian coordinates, we have x=[ϕvs]T, S=∫abρ0sdη, δxS=[00ρ0]T. The underlying matrices are P0=02×2, P1=[0110], G0=02×1, G1=[01]T, C(x)=0, f0=f=ρ01, R1=ρ0Tμ∂ηv, R0=0, gs=ρ01, rs=ρ0T2k∂ηT, γ1=ρ02Tμ>0, γ0=0, γs=ρ02T2k>0. Considering the state variables as x=[ϕvs]T, the energy balance in Lagrangian coordinates can be expressed as follows
where σs:=ϕρ0Tμ(∂ηρδvH)2+ϕρ0T2k(∂ηρδsH)2≥0 is the internal entropy production, ysTus:=(Tq)ab. □
Conclusion
After recalling the constitutive relations of non-isentropic fluids in both Eulerian and Lagrangian coordinates, we show how these systems can be cast as Irreversible Port-Hamiltonian Systems by extending the definition in Ramirez et al. [9] to include convective transport. The key modification concerns the differential operators, whose structure introduces convective terms that must be treated appropriately. As in Ramirez et al. [9], we then derive a parametrisation of the boundary port variables that ensures consistency with the first and second laws of thermodynamics.
References
[1]Araque, L.M. (2020). Port-Hamiltonian modeling of fluid-structure interactions in a longitudinal domain. Ph.D. thesis, Université Bourgogne Franche-Comté; Universidad técnica Federico Santa María . . . .
[2]Cardoso-Ribeiro, F.L., Haine, G., Le Gorrec, Y., Matignon, D., and Ramirez, H. (2024). Port-Hamiltonian formulations for the modeling, simulation and control of fluids. Computers & Fluids, 283, 106407.
[3]Favache, A., Dochain, D., and Maschke, B. (2010). An entropy-based formulation of irreversible processes based on contact structures. Chemical Engineering Science, 65, 5204–5216.DOI
[4]Grmela, M. and Öttinger, H. (1997). Dynamics and thermodynamics of complex fluids. I. development of a general formalism. Physical Review E, 56(6), 6620–6632.DOI
[5]Hoang, H., Couenne, F., Jallut, C., and Le Gorrec, Y. (2012). Lyapunov-based control of non isothermal continuous stirred tank reactors using irreversible thermodynamics. Journal of Process Control, 22(2), 412–422.DOI
[6]L.A. Mora, Y.L.G. and Ramirez, H. (2024). Energy-shaping and entropy-assignment boundary control of the heat equation. Systems and Control Letters, 189, 105821.DOI
[7]Malvern, L.E. (1969). Introduction to the Mechanics of a Continuous Medium. Monograph. Prentice Hall, Inc.
[8]Mora, L.A., Le Gorrec, Y., Matignon, D., Ramirez, H., and Yuz, J.J. (2021). On port-Hamiltonian formulations of 3-dimensional compressible Newtonian fluids. Physics of Fluids, 33(11), 117117.DOI
[9]Ramirez, H., Gorrec, Y.L., and Maschke, B. (2022). Boundary controlled irreversible port-hamiltonian systems. Chemical Engineering Science, 248, 117107.DOI
[10]Ramirez, H. and Le Gorrec, Y. (2022). An overview on irreversible port-hamiltonian systems. Entropy, 10, 1478.DOI
[11]Ramirez, H., Maschke, B., and Sbarbaro, D. (2013a). Irreversible port-Hamiltonian systems: A general formulation of irreversible processes with application to the CSTR. Chemical Engineering Science, 89(0), 223 – 234.DOI
[13]Ramirez, H., Le Gorrec, Y., Maschke, B., and Couenne, F. (2016). On the passivity based control of irreversible processes: A port-Hamiltonian approach. Automatica, 64, 105 – 111.DOI