Dynamic plant simulations

A plant simulation is the set of equations necessary to approximate the response of a chemical plant to various changes. A steady- state plant simulation is one that predicts the eventual outputs when the inputs and all the internal variables are held constant. It does not say how the outputs are reached. A dynamic plant simulation is one that predicts how the outputs of a plant will change when a known change in the input occurs. It gives the path the process follows in going from one steady state to another. 

Figure 2.3 Applications of steady state and dynamic Plant Simulation Models...

Many HVAC system engineering problems focus on the operation and the control of the system. In many cases, the optimization of the system s control and operation is the objective of the simulation. Therefore, the appropriate modeling of the controllers and the selected control strategies are of crucial importance in the simulation. Once the system is correctly set up, the use of simulation tools is very helpful when dealing with such problems. Dynamic system operation is often approximated by series of quasi-steady-state operating conditions, provided that the time step of the simulation is large compared to the dynamic response time of the HVAC equipment. However, for dynamic systems and plant simulation and, most important, for the realistic simulation... 

To develop integrated thermal-hydraulic and chemical plant models and plant simulators to investigate the dynamic behaviour of the plant. 

McAvoy (1999) advanced the use of optimization calculations at the controller design stage, proposing the synthesis of plant-wide control structures that ensure minimal actuator movements. The initial work relying on steady-state models (McAvoy 1999) was recast into a controller synthesis procedure based on linear dynamic plant models (Chen and McAvoy 2003, Chen et al. 2004), whereby the performance of the generated plant-wide control structures was evaluated through dynamic simulations. 

The plant simulation considers only a reduced number of units, but dynamically representative, as follows. Crude EDC from R1 and R3 are sent to washing/drying in the unit SO. Dissolved gases and very light impurities are removed in SI, and further in the distillation column S4, which is the exit point of the light impurities. After pretreatment, the crude EDC is sent to purification in the distillation column S2, which is the key unit of the separation system. This column receives crude EDC from three reactors. It is also the place where three large recycle loops cross. The top distillate of S2 should remove the light impurities mentioned above, while the purification of EDC from heavies is continued in the distillation columns S3 and S5. 

Unsteady-state or dynamic simulation accounts for process transients, from an initial state to a final state. Dynamic models for complex chemical processes typically consist of large systems of ordinary differential equations and algebraic equations. Therefore, dynamic process simulation is computationally intensive. Dynamic simulators typically contain three units (i) thermodynamic and physical properties packages, (ii) unit operation models, (hi) numerical solvers. Dynamic simulation is used for batch process design and development, control strategy development, control system check-out, the optimization of plant operations, process reliability/availability/safety studies, process improvement, process start-up and shutdown. There are countless dynamic process simulators available on the market. One of them has the commercial name Hysis [2.3]. 

Hence, the development of a Plant Simulation Model is the proper approach to deal with industrial simulation problems. The progress in software technology makes possible today the development of integrated steady state and dynamic models. However, these require significant investment in qualified staff. Recently, generic simulation products have been proposed for applications in refining and petrochemical industries, which can be customised for specific processes. 

The job is not finished with steady state controllability analysis. Only dynamic simulation enables a reliable assessment of the control problem. The solution of the dynamic modelling depends on the dynamics of units involved in the control problem. Detailed models are necessary for the key units. The simplification of the steady-state plant simulation model to a tractable dynamic model, but still able to represent the relevant dynamics of the actual problem, is a practical alternative. Steady-state models can be used for fast units, as heat exchangers, or even chemical reactors with low inventory. 

A steady-state Plant Simulation Model of an existing plant helped to calibrate the base-case model on a representative operating point. Some details of an industrial process were skipped, but the omission of these details does influence neither the plantwide material balance nor the process dynamics. The units SO, S6 and S7 may be considered black-boxes. Contrary, SI to S5 are rigorous distillation columns, modelled as sieve trays. In steady-state all the reactors are described by stoichiometric approach, but kinetic models are used for Rl and R4 in dynamic simulation. 

In validation process, the semi-dynamic simulation tests were also performed additionally to investigate the integrity for system requirement. By use of the plant simulators, their response to the transients were verified such as LOCA, LOP A, and Main Steam Isolation Valve Closure etc. 

Given the design decision to use < > = 0.25, based on the steady-state C R analysis, verification is performed by dynamic simulations with HYSYS.Plant. The hot stream of n-octane at 2,350 Ibmol/hr is cooled from 500 to 300°F using n-decane as the coolant, with F2 = 3,070 Ibmol/hr and F3 = 1,200 Ibmol/hr. Note that these species and flow rates are chosen to match the heat-capacity flow rates defined by McAvoy (1983), with F, slightly increased to avoid temperature crossovers in the heat exchangers due to temperature variations in the heat capacities. Additional details of the HYSYS.Plant simulation are... 

Because of the computational complexities associated with dynamic process simulation for rrmltiunit processes, there is still much to be done before simulators of this type become available for general application. Another problem complicating their development is that process nnodels for even individual separation units ate usually for steady-state cases this is the result of both incomplete understanding of the chemical and physical nciples involved and computational difficulties. This is one of the main reasons why process control considerations ate difficult to incorporate into chemical process simulation and thesis and why on-line plant optimization is still far away in most instances. 

We compared our algorithm GSPAR with FRONTAL, the frontal method of SPEEDUP, using four example matrices arising from dynamic process simulation of chemical plants. The computation time (in CPU sec.) for a Cray Y-MP is given in table 4.1. 

In the proposed framework, HYSYS. PLANT simulation package is used to validate both the steady state and dynamic models even though the switchability fiom steady state to dynamic mode is not a trivial procedure, as it will be shown in the case study section. 

The goal of plantwide control structure synthesis is to develop feasible control structures that address the objectives of the entire chemical plant and account for the interactions associated with complex recycle and heat integration schemes, and the expected multivariate nature of the plant. Many strategies have been proposed for accomplishing this task, and the majority of them have been demonstrated using dynamic process simulations. However, none have been accepted as the universal approach, in a manner similar to the steady-state process design synthesis hierarchy of Douglas [1]. 

Computer simulation of plant operation may also be made offline real time using computer models of the complete installation, which are capable of simulating dynamic plant response to changes in operating parameters, plant upsets, etc. Such systems may be used for off line optimization studies and for operator training in handling emegencies, start-up- and shut-down situations, etc. Without risk to plant or personnel. Simulators are described in [751-758, 923]. 

The main pressure pipe break accident analysis was conducted using the French OASIS Code, which is a dynamic system simulation program especially for the pool type sodium cooled FBR. It can simulate the thermal-hydraulics of the whole FBR plant circuits and reactor control and protection system, including the regulation system. So OASIS is a good simulation tool to study the operation and accident transients in a FBR Plant. An introduction and the physical models of the OASIS code are presented in the Ref. [1]. 

Mathematically speaking, a process simulation model consists of a set of variables (stream flows, stream conditions and compositions, conditions of process equipment, etc) that can be equalities and inequalities. Simulation of steady-state processes assume that the values of all the variables are independent of time a mathematical model results in a set of algebraic equations. If, on the other hand, many of the variables were to be time dependent (m the case of simulation of batch processes, shutdowns and startups of plants, dynamic response to disturbances in a plant, etc), then the mathematical model would consist of a set of differential equations or a mixed set of differential and algebraic equations. 

---