How can I control fluid library objects with the fluid component names?

Hi,

The fluid library components are well thought out, but not all of the underlying capability is surfaced, thus application is severely curtailed.

The greatest limitation lies in the inability to control the process with the fluid component names entered into the ticker, thereby losing the capability to simulate substreams with conserved properties.

Currently the paradigm is to control everything based on streams of discrete fluid from each port, so that I pull a certain volume from port 1, and from port 2 etc. I can then mix them, split them or delay them mediated by port volume, total volume etc. I can also fiddle with these numbers dynamically through the triggers. This is a given.

But I can't seem to model a situation where a generator produces a stream of two components, at a certain percentage on a mass basis, and the downstream objects, such as pipes, mixers, reservoirs, processors and splitters, can keep track of these component rates and take actions based on them. In this situation, the rate of material and the percentage of each component generated will then vary over time according to some input signal.

The main value of Flexsim, from a holistic perspective, is derived through its information conservation properties that operate through its calculations processes. This is well-expressed in the discrete library, where one can have flow items with their own properties, and process them to some conclusion, calculating the changes in those properties along the way. Thus, engineers can use Flexsim as a simplified physical process simulator, and solve some complex problems using the DEVS approach, such as desorption from a solid over time mediated by particle diameter. Essentially, Flexsim becomes then a sophisticated DEVS calculator, summing various results over time.

But the fluids library lacks this information conservation foundation, and thus one cannot model fluids with the same level of sophistication as the discrete library.

At present I can't seem to be able to:
- keep track of the indivudal components in reservoirs, pipes
- use the components of any stream as the basis for any operation
- split streams based on component, rather than just volume
- mix A+2B -> C+2D, with mass conservation

Ideally, Flexsim would enhance the current library so that either approach (Volume or component) could be used, thereby opening the door to simulation of common industrial processes.

With the small addition of property labels to each fluid component, constant or variable (e.g. molecular weight, viscosity), and conditions labels to each operation (e.g. temperature, pressure), one could setup a simple, chemical process simulation where the fluid properties change according to conditions and reactions (i.e. mass conserved, single step, mixing function where mass ratios of input components are added into mass ratios of multiple output components).

These changes would enable Flexsim to be used for industrial fluid simulations, where Flexsim can be used as DEVS calculator for chemical processes. I would argue that the framework for this capability is already in place, so it seems frustrating that it is not surfaced.

Short of Flexsim surfacing the latent functionality of the fluid components, is there any way that I can code the objects to act according to the fluid components?

Please help

Thanks

Brett,

We had much of the similar issues as you described. With the current objects it is hard to model a flow in Flexsim that is both accurate and fast in execution. The current Flexsim paradigm is using time slicing and it is only possible to pass a 'parcel' of product to the next downstream object within one tick time. If the next object cannot hold more than the [flowrate * time slice period] you have an issue and you need to reduce the time slice duration. Also diverting a % of a flow upstream will give you headaches in the current library.

We (Talumis) have created a Flexsim library and calculation method that uses a complete new way of flows modeling. It is possible use splitting on %, having preferred flows, delays in flow and tanks usage with accurate and extremely fast calculations .We are able to accurately trigger on tanktops/bottoms, any number of tank trigger levels in our mechanism and even triggering when a specific volume has been sent out. This library is ideal for ‘logistic’ analyses of chemical (oil/gas) plants and supply chains. We are currently busy in creating a commercial library for other Flexsim users. So if you (or anyone else) is interested, let me know.

Best regards,
Dirk-Jan Moens
TALUMIS

Hi Dirk,

You are correct in that one clear need for the fluid library is the capability to model real-world items that include complex flow networks. But this observation must be tempered with the fact that there is a slippery slope to this type of analysis, and Flexsim must only develop the capability that is needed to cohere to its vision and market position.

In my view, the limits of the type of analytical solving Flexsim should attempt, is to be able to accurately calculate the distribution through a complex network with a number of different inlets and outlets. I imagine the idea of your logistics network of fluid objects has great appeal to many Flexsim users, as it seems like a good one to me.

However, my concern is still the inability to have information conservation through the system (i.e. where the fluids are comprised of components, each with properties), so that the components are tracked and their values can be used in operations (e.g. one component of the stream is inert).

In my situation I have >200 vertical gas wells, spread out over a 30km x 15km area producing different flow rates of:
- Methane
- Carbon Dioxide
- Oxygen
- Nitrogen
- Ethane
- Higher Hydrocarbons (propane, butane ...)
- Water

Ideally, I would model the situation as a separate generator and network for both gases and water, since this is pretty close to what happens on site. Since I am using a mass basis of kg/hr for fluids, Flexsim should be able to generate gas production every tick, using a set percentage of the different fluids and a total mass flow rate. This gas component information, set using the Initial Product ID button, in the generator properties box, should be conserved throughout the flow network, and one should be able to use it in operations (e.g. mix component 1 from port 1 with component 2 from port 2, to create product 3 and product 4 in a mixed stream of known ratio).

Instead I have to separate all of the flows into individual pure streams, and so I need an additional 1,000 extra fluid generators, and a few thousand extra virtual pipe connections, all invisible, simply to simulate a simple combustion process, which requires information conservation.

Thus, to me, the fluid library seems to lack the conceptual depth and well-refined nature of the discrete library, and is at a more rudimentary, less evolved stage, with a lot of untapped potential.

I would be very interested to hear from the Product Manager, or official team on their plans for the fluid library.

Brett,
We have dealt with component tracking/splitting/mixing for LNG supply chains. So the components you describe are well known. I am wondering tough what kind of logistics problem you try to solve here.
In any case there will not be a library available for third party use before the end of the year. I will keep you posted on progress.
I was wondering with the kind of model you are creating do you not have speed issues? And what time dependent dynamics do you need to consider?

Rgds,
Dirk-Jan