Understanding SimSinter

Steady State Simulation

The majority of Simulations run on SimSinter are Steady State simulations. Steady State simulations do not have a time component. They simply simulate the simple ideal case of the reactor running at equilibrium according to the provided constants and input variables. Aspen Plus and Excel can only do Steady State simulation. ACM supports both Steady State and Dynamic simulation. gPROMS supports both Steady State and Dynamic simulation, but SimSinter can currently only perform Steady State simulations with gPROMS.

Steady State simulations have a single set of inputs and outputs. SimSinter sets the inputs before the simulation starts, and collects the outputs when the simulation completes and returns them in the output file.

Dynamic Simulation

Overview

Dynamic simulation is much more complex because it involves time. Dynamic simulation is used to see how the reactor responds to changes over time, and can be used to simulate conditions such as starting up, shutting down, faults in the system, changes in fuel mix, etc. The most important component of the Dynamic simulation is therefore the “TimeSeries” which lists the time steps the simulation will go through.

Dynamic simulation is currently only supported by ACM, and the Dynamic simulation features were designed to meet the needs of the DR-M builder. The dynamic features are general enough that other projects may find the dynamic simulator useful, but more development may be required. Please send an
e-mail to ccsi-support@acceleratecarboncapture.org with any requests for improvements or bugs.

TimeSeries

The Dynamic simulation moves through time in accordance with the TimeSeries. The TimeSeries is an array of doubles, where each double represents the time that the time step will END (and the next one will begin). The dynamic output variables are read from the simulator at these breaks, and the input variables are set.

Time

S imulation Start

T imeSeries Time 1

T imeSeries Time 2

T imeSeries Time 3

T imeSeries Time 4

I nput

Input 1 set

Input 2 set

Input 3 set

Input 4 set

–

Ou tput

–

Output 1 read

Output 2 read

Output 3 read

Output 4 read

Snapshot

Snapshots are a feature of ACM that allow Dynamic simulations to be restarted from a saved time and condition. For example, a user may want to run through a fault scenario multiple times with slightly different starting conditions. Saving a snapshot just before the fault scenario allows this to be done efficiently.

To use snapshots from SimSinter, a user must be careful to do three things correctly:

  1. Set the Snapshot “input” setting to the name of the snapshot to start from.

  2. Set the “TimeSeries” such that the first value is strictly greater than the snapshot time, and the values monotonically.

  3. The simulation must be distributed with the “AM_???” subdirectory created by ACM, containing any .bak files found there. ACM stores the snapshots in those .bak files, so if they are not included, ACM will not be able to restore the snapshot.

Variables in Dynamic Simulations

Dynamic simulations have four kinds of variables.

  • Steady State Input Variables are functionally equivalent to the input variables of Steady State simulations. They have a single input value that is set at the beginning of the run and is never changed. Actually, that value is reinserted at each time step break, so if the value changes in the simulation it will be reset back to the initial value at every time step.

  • Steady State Output Variables are functionally equivalent to the output variables of Steady State simulations. Only the value found at the last time step is returned in the output data. Steady State output variables are mostly only useful for statistical data in Dynamic simulations.

  • Dynamic Input Variables have values that change at each time step. Internally Dynamic input variables are represented in SimSinter by an array of the same length as the TimeSeries array. At each time step the input variable in the simulation is updated to the value ad that address in the array. Currently there is no fine control of ramping the values up and down included in SimSinter. ACM has internal controls for handing the change in input values that may be modified by the user. Dynamic input variables also only have a single default value for the whole input array. Therefore, a dynamic scalar only has a single default value (e.g., 5) although the input data is represented as an array. As a result, if the user does not provide a Dynamic input variable (as an array) in the input data, that variable will hold its single default value throughout the run as if it was a Steady State variable.

  • Dynamic Output Variables return the value of the simulation variable at the end of each time step. Internally Dynamic output variables are represented in SimSinter by an array of the same length as the TimeSeries array. At the end of each time step the output variable in the simulation is read and entered into the correct address in the array of the Dynamic output variable in SimSinter.

Data Layout

The most confusing thing about Dynamic simulation is how the various pieces of data are split between the configuration file and the input file in practice.

  • TimeSeries – The configuration file may contain a default TimeSeries as a 1 dimensional (1-D) vector of doubles. However, it does not have to. DR-M Builder never uses the default TimeSeries, so in most cases a valid TimeSeries default does not need to be provided in the configuration file.

  • Snapshot – Similarly, the configuration file also contains a default snapshot name. This can also be overridden in the input file, but DR-M builder rarely bothers to use Snapshots at all.

  • Dynamic Input Variables – In the configuration file Dynamic input variables have their own section (dynamic-inputs) but otherwise look the same as normal inputs, they have the same data and meta-data in the same layout. The default of a dynamic scalar double is just a single double, not a vector. In the input file there are no separate sections for Dynamic and Steady State variables, they are all in the same section. However, Dynamic input variables have the time dimension, so a dynamic scalar is represented as an array of doubles in the input file, and a dynamic vector is represented as a 2-D matrix of doubles.

  • Dynamic Output Variables – In the configuration file Dynamic output variables have their own section (dynamic-outputs) but otherwise look the same as normal outputs. They have the same data and meta-data in the same layout. Dynamic outputs do not appear at all in the input file, but in the output file there is no separation between Dynamic and Steady State variables, they are all in the same section. However, Dynamic output variables have the time dimension, so a dynamic scalar is represented as an array of doubles in the output file, and a dynamic vector is represented as a 2-D matrix of doubles.