How to choose between PowerSim 8 Academic and the Academic SDK

Hello to everyone reading this. If you have landed on this page, you are most likely looking for practical information about PowerSim 8. I know the feeling, because I was in exactly the same position myself, and unfortunately searching the Internet never turned up the answers I was hoping for.

So, down to business. The research institute I work for bought PowerSim 8 SDK to build and test various dynamic economic models.
I will admit that we had been working with PowerSim 2.5 and were perfectly happy with it, until we built a model on such a scale that 2.5 threw an error about exceeding the permitted number of variables. Some careful surgery on the dll (there was a fair bit of voodoo involved) got that limit lifted. The economists had barely finished celebrating the miracle when PowerSim started glitching. You would be building a model and then, out of nowhere, a memory error came up and the environment shut down (without saving, of course). There was clearly no way to carry on working like that.
Then one day we were given funding to buy a licensed PowerSim 8. The money was limited, naturally, so our choice had to be limited with it.
After reading through the descriptions on PowerSim.com we settled on the academic version of the product together with the academic SDK. And that is where it starts to get interesting!
A word of warning! When you pick an academic license, the site shows you the table below (prices as of 14.10.2011):

Prices for Studio 8 Academic products

It all looks simple enough. We went for the Academic Standard Bundle + SDK. That was our fatal mistake!
The problem is that the configuration we picked does not include the FeaturePack add-on bundle.
Without the add-ons you get the most cut-down version of PowerSim there is, missing most of the convenient functions (the mathematical ones included).

Take this formula: (TotalOborot-TotalOborotReal)
TotalOborot holds a range of values. We could of course set that range by hand, like this: ({0.1,0.5,0.9}-{0.444,2.33,2}), and that would be fine as long as the model only had a handful of variables.
Now picture a model with more than 2000 variables (ours is one of those). On top of that, the value ranges have to be entered by the user. PowerSim has a function for exactly this job: XLDATA(“file.xls”;”SheetName”;”A1:A10″), which pulls data out of an xls or xlsx file.
So you would expect everything to work nicely once you write the formula: (XLDATA(“file.xls”;”SheetName”;”A1:A10″)-XLDATA(“file.xls”;”SheetName”;”B1:B10″)). Instead you get an error: ‘XLData’ cannot be used in composite expressions. In other words, you cannot nest the function.
The way around this is the XLRANGE() function, which can be nested, BUT IT SHIPS WITH FeaturePack 3. We could not use it, because we had not bought the FeaturePack. We had no choice, so we worked around it by creating extra Auxiliary elements, each one holding an xldata(); call. We then used those Auxiliary elements as variables. That gave us the main formula back in the form (TotalOborot-TotalOborotReal).
The approach works, but it “inflates” the number of variables n-fold.
Conclusion: do not be cheap, buy PowerSim Academic Standard with Feature Pack.

Now for the PowerSim 8 Academic SDK.
Here is the answer to the question I kept asking myself before we bought it: “what can this SDK actually do?”. There turned out to be very little information about it, either on the official site or anywhere else on the Internet.
What I most wanted to know was whether you can create a model without the visual editor. That is, whether you can create a model file programmatically and then edit an existing one.
Well, the SDK gives you nothing of the sort!
Through the SDK you can work with a model at the level of reading data. You can also drive the PowerSim engine to calculate with the data you give it, but changing the model itself (a formula, for instance) is IMPOSSIBLE!.
At least, I could not do it. If anyone has managed it, please tell me how.

Michael Dvornichenko Solution architect · SK.AI