Tutorial 1
Getting started

 

Learning goals

  1. Getting familiar with Quaestor interface.
  2. Adding parameters, relations and constraints.
  3. Providing references and properties for parameters.
  4. Creating a multiple case solution.

Prior knowledge

  1. Interface

1    Objective

As mentioned earlier these tutorials are focussing on providing the basic skills of knowledge engineering. In this first tutorial, the main dimensions of a ship will be systematically varied. The knowledge used is Archimedes’ Law in combination with design requirements, which are the following:

The ship should have

Our objective is to generate a dataset of systematically varied ships that meet the design requirements.

2    Before you start

Before starting with your knowledge engineering, please make sure you are actually using the knowledge engineering user level and you have Quaestor configured in the most convenient layout.

To check/change your user level go to Tools -> Options -> General. You can change your User level. If you are not able to change it to Knowledge Engineer (level 3), you do not have KE rights. Please contact your ICT department or Qnowledge for the proper license.

3    Creating parameters

When Quaestor is started, you are presented with the opening screen, with no knowledge base or project loaded:

 

To create your first knowledge base

The screen changes to the main Quaestor knowledge engineering screen:

When a knowledgebase node is expanded, like Newqkb here, the classes within the knowledgebase are visible. Classes are used to organise your data and equations, and put them in a practicable structure. In Quaestor, all data is put in parameters. You can designate one (or more) parameter as the goal of the calculation.

Classes are a means for organizing parameters for the user. They are not to be confused with a class in object oriented computer programming. The Quaestor core does not use the classes at all.It only uses the parameters!

The Top Goals/Undefined class is part of the knowledge base by default. Here, the goal parameters of your calculation (Top Goals) can be stored, as well as other parameters that don't belong to other classes (Undefined).

Create a new class, named Main Dimensions, in which the following ship parameters are to be stored.

Lpp: Length between perpendiculars
B: Moulded breadth
T: Draught from keel to Construction Water Line

Sub classes can be defined by right clicking on any class name and again selecting New Class.

The class is now shown in the knowledgebase tree, and you can add the parameters Lpp, B and T.

The added parameters now appear in the right part of the Knowledge browser like this:

A red cross draws your attention to the fact that a parameter is not yet fully defined for the system to use it.

In Quaestor, a valid parameter has:

  • A unique name;
  • A dimension (or is assigned dimensionless);
  • A way to be determined: input from the user or defined by a relation.

The first requirement is already satisfied, so let's assign dimensions to the parameters.

You can also add a reference for a parameter. A proper reference is important, as users of the knowledgebase may not be certain of the exact meaning of the parameter. Please also read Documentation of knowledge for more details.

Now only the last requirement for valid parameters has yet to be fulfilled: a way to be determined (input from the user or defined by a relation). This qualification can be provided in the Properties window.


Notice that now, the status of the Lpp parameter has changed to

to show that Quaestor knows enough about this parameter to use it in a computational model. Now make sure that B and T are determined by VR: User only too. By the way, VR stands for Value Requested.

If a parameter still shows a red cross when you have provided all needed property values, click on Top Goals/Undefined, and you will see the blue cross.

Note that above you have explicitly added Lpp, B and T as parameters to the knowledge base. You are also able to implicitly add parameters by creating relations that contain new parameters. Quaestor will automatically add these new parameters to your knowledge base (so, implicitly). You will encounter this in the following section.

4    Creating relations

In order to define the loading capacity of the ship, it’s important to know the kind of water (Salt or Fresh) the ship will sail in. Therefore, you're going to add some relations and constraints, and make sure the user can select the watertype. Two relations must be added, one for salt water and one for fresh water.

Rho = 1025

(belonging to salt water), and click Save:

Note that when you enter a relation, Quaestor provides as much help as possible by means of the Help Checker. This checker shows what to expect (in this case a Value or an Expression, ValExp). After saving a relation, the Help Checker will check the syntax for possible errors, and shows a warning message when something is wrong.

Because of entering the relation, the parameter Rho has automatically been created. For Rho to be a valid parameter, a dimension should be assigned.

The parameter is automatically created in the Class in focus when saving the relation!

What follows now might seem a bit unconventional: a second relation is given for Rho. This is a nice example of how Quaestor works: in a calculation, Quaestor can find numerous calculation paths and automatically chooses the most appropiate one, based on constraints and available data.

An infinite number of relations can be assigned to one particular parameter.

Rho = 998

5    Creating constraints

The two relations for the density of the water are only valid for their corresponding watertype. Therefore, a constraint is added to each relation. A constraint is simply a restriction for the validity of the relation.

Click Top Goals/Undefined in the Knowledge browser: all relations in that class are shown (this will work for any class). One can always edit a relation by selecting it and pressing F2 (or right click and select Edit). This is also a useful method to find out which relation you actually selected.

Watertype$ = "SW"

and click Save:

The constraint is now added, and a parameter named Watertype$ has been created. The $ at the end of the parameter name makes sure Quaestor recognizes it as a String value, and the dimension Str is automatically assigned (see also Quaestor syntax). You have to assign a Determined-by value to make it a valid parameter.

Repeat the process for the other relation.

Watertype$ = "FW"

to the relation Rho = 998.

Note that the expression editor assists you with the presentation of existing parameters and intrinsic functions. 

Both relations are now connected to the corresponding watertype by means of the value of Watertype$.

Whenever Rho is needed in the calculation process, Quaestor will note that the watertype is needed to determine the value of Rho, and will ask the user for this value (because it is assigned to be User only).

When you select a parameter in the upper list, the relations of that particular parameter are visible:


Moreover, when you select a relation in the lower list, the constraints for this relationship are shown:

6    Creating a dropdown menu

Because there are only two possible values for Watertype$ (SW or FW), it's easy to integrate a dropdown box. 

FW<EQ>Fresh Water

SW<EQ>Salt Water

The use of a dropdown box will be clear in a later stage of this tutorial, when calculations are made. Note that you can add the @EQEXPLAIN attribute in the Data box ('data slot') of the Watertype$ parameter in order to display only "Salt Water" and "Fresh Water" are shown in the dropbox (so without SW and FW).

7    Creating a relation for the displacement

On the basis of the earlier defined parameters and the block coefficient Cb , the displacement of the ship can be calculated.

DISP = Cb * Lpp * B * T * Rho / 1000

which defines the displacement of the ship in tons.

Note that two new parameters are created: the block coefficient Cb and the displacement DISP. In order to keep the knowledgebase meaningful, it is necessary to provide references, dimensions and determined-by values for these parameters.

8    Creating a solution

Let's perform your first Quaestor calculation. A solution is always determined by one or more Top Goals. A top goal is a parameter (or object) that is your final calculation target, in this tutorial it's the displacement of the ship.

Note that the cross in front of DISP changed to , indicating that it's a top goal for the calculation.

Calculations and solutions are managed in the Workbase. Here, solutions can be created, redone (with different data), examined and deleted. 

The calculation progress is started, and Quaestor first wants to collect all input data in the Workbase window:

Entering data is easy: just type a value for each selected parameter and press enter to switch to the next one.

B = 10   Cb = 0.55   Lpp = 60   T = 6  Watertype$ = SW (Salt Water)  

Note that the input of Watertype$ consists of a dropdownbox, which you created above:

After providing the data you have to press the Next button to continue. The actual calculation is started now and because this is a very simple example, the final state is almost immediately shown:

The top goal (DISP) is shown, together with all parameters that were used for the calculation. With these input values, the displacement of the ship is about 2030 tons.

9    Multiple case solutions

So far, you created one solution, which is shown in the workbase:

As mentioned in the objective, you will create a dataset of systematically varied ships. You could of course perform several calculations with different input data by hand, but it's much easier to use the ability of Quaestor to create multiple case solutions. To keep the complexity of this example within reason, you will only vary the breadth and length of the vessel. Please note that Quaestor can only perform calculations on ranges of parameters when you already have created a solution.

 

B= 9(0.5)11    Lpp= 55(2)69

In this way, B and Lpp are defined by a range, given the start value, step size and end value. For example, Lpp is defined from 55 to 69 meter, with steps of 2 meter.

Quaestor will ask if it should create a case matrix for Lpp (figure 15), click Yes (an explanation will follow).

You will see that the single value input is still in the list and the multi case values are in the table part of the Workbase. Press the Next button to continue.

The new solution is now created. The fixed values are again shown in the right hand side of the workbase. The varying parameters (B and Lpp) and the corresponding solution for DISP are shown in the lower part. Each row is a different case, identified by a case number(#1, #2, etc.).

What does the Case matrix question mean?  Note that the variations of B do not correspond to the variations of Lpp: there are 5 values for B and 8 values for Lpp. If you would have answered No to the Case matrix question, five variations of B would have been made, each one with the corresponding value of Lpp (the last 3 Lpp values would be omitted). In that way, there would be only five cases. As you answered Yes to the question, all possible combinations are considered, which results in 5*8=40 cases. You can easily notice the difference by creating a new solution with the same input values, but now answer no to the case matrix question.

Ranges can be defined in multiple ways: 

  • Individual values can be separated by comma’s: 9, 10, 11 or by semicolons. This depends on the setting of Tools->Options->Modeler->Use ; instead of , in range input.
  • You can define a range with a begin, an end and number of steps (intervals): 9(&2)11
  • You can define a range with a begin, an end and a step size: 9(1)11
  • You can define a range with a begin, an end and a step size, including the last element when this is smaller than the whole amount of steps: 9(#1)11.5

10    Filter results


As it is, the multiple case solution table is sorted by the breadth of the ship. Now suppose you'd like parameter Lpp to be the leading parameter in this table.

A lot of cases in your solution don't meet the design criteria: a displacement between 2000 and 3000 ton. We could have fixed this by adding a constraint to DISP, but another way is to use a filter in the solution table.

Now, only the cases for which 2000 < DISP < 3000 are shown in the table. The objective of this tutorial is now completed: you built a dataset of systematically varied ships that meet the design criteria.

11    Check

You can verify your results by comparing it to  [Tutorial 1 Finished] .

Continue to tutorial 2 >>