Figure 1. A typical courses dependency graph - dependencies for Computer Architecture
SIMUNIC, Juraj Ph.D. & DUROVIC, Zdravko B.Sc.
University of Rijeka - Faculty of Engineering, Vukovarska 58, HR-51000 Rijeka, Croatia, zdjuro@rijeka.riteh.hr
Abstract: Issues of undergraduate level curriculum design are analysed. A semiformal design process is explained as used for the design of the first full-time undergraduate nine-semester curriculum in electrical engineering at the University of Rijeka.
The design process separates the following phases:
and each phase and its constituent steps are described in considerable detail. The process incorporates means for designing a new curriculum and associated documents in a controlled and predictable manner - a necessity if a high-quality set of documents has to be built by a distributed group of around fifty authors confronted with a hard deadline. The merits of the highly iterative design process are described.
Only partial conclusions could be given, due to the fact that the only true test of any curriculum (and, in consequence, of its design process) is its execution. Therefore, complete conclusions will have to be made in five or six years from its inception this autumn, after the first generation of the students graduates.
Keywords: curriculum design, semiformal design process
Designing a curriculum in any mature field is a demanding task, even when the milieu of such endeavor is completely favorable.
New curriculum is usually built upon the old one, based on the experience gained through many years of putting the old curriculum into educational practice and patiently noting its flaws and virtues. Field's university tradition provides the staff needed for writing of the new curriculum and, later, implementing it, using mostly space and equipment that are already available. Ideally, there would be an infinite amount of time for writing, consultations, reviewing, proofreading and various preparations, and in reality, while being far from infinite, the time is still plenty, in part due to the fact that the old curriculum can almost always be used for another year or two.
But, designing the first curriculum in a given field at particular university, especially in a fast changing field like electrical engineering, presents the designers with an inflated set of problems. Authors have been in this less common situation approximately for the last three years of writing and rewriting an electrical engineering curriculum, planned to become the first four-year undergraduate electrical engineering course at University of Rijeka. This paper describes more prominent of the considerations, choices, trade-offs and decisions made by the authors and their colleagues in the course of designing the curriculum, together with the design process.
When taking into consideration a highly influential document like a curriculum, we have to think about the environment into which it is placed, for two main reasons:
Environment of a curriculum can be seen as internal to the university (which is visible to students during their studies), or external (which becomes visible after graduation), and we can talk about separate university and industrial environments of a curriculum.
Engineering education at the University of Rijeka has a long tradition in machine, naval and civil engineering. University represents focal point of education for the three counties that form Primorje region with around 400000 people.
Most of the engineering education at the University is done within the Faculty of Engineering, whose staff carries out nine-semester courses in mechanical engineering and naval architecture, and five-semester courses in mechanical, naval and electrical engineering.
University-level education in Croatia is going through administrative changes, in which (roughly) all four-year and longer courses will remain at the universities, and all shorter courses will be moved to newly instituted polytechnics. Since the process of separation has just begun, university staff still carries out all courses.
Electrical engineering education is a relative newcomer to the University, with a short five-semester course in electrical power engineering yielding around 30 engineers per year. Shortness of their education results in its narrowness, becoming an impediment in their effort to become practicing engineers. In consequence, due to the changes in local economy and saturation of the local market for electrical power engineers, many of them indeed find employment outside electrical (or any other) engineering.
Most of the local industry is centered in the coastal cities. Traditionally, the region was oriented towards large metal-working industry, construction, oil and chemical industry, electric power generation and transmission, food production, tourism and transportation. Wars and recent recession have seriously weakened metal-working, construction and transportation, and large companies (with a few exceptions) are not thriving. Since there has never been any large electrical industry at all, electrical engineering is primarily seen as a support, and electrical engineering companies are application (rather than production) oriented.
Even in the current economy crisis, local human resources market raises its demands for qualified electrical engineers to the point where there are no unemployed graduate electrical engineers at all, and supply (mainly local people who studied at some other university) can't even come close to the demand. Croatian system sharply demarcates engineers (five or six semester courses) and graduate engineers (four year and longer courses), so it is possible to have a surplus of a particular kind of engineers, and the lack of similar kind of graduate engineers.
Unfortunately, complete cure for the problem, in the form of opening a complete electrical engineering university education with enough courses to cover the entire field, is impossible to administer. On one side, there is an economy crisis in progress, making any new investments questionable - and starting any engineering education represents a substantial investment. Complete electrical engineering programme would simply be unbearable for the local economy resources. On the other side, immense complexity of a complete programme makes any effort at starting it from scratch unlikely to succeed.
Thus, the new electrical engineering curriculum has to be targeted to those subfields within electrical engineering that are needed most, and that will still be needed when students start to graduate it some five or six years from now.
Electrical engineering is a vast field, and many partitions into subfields can be proposed, ranging from extremely coarse (like the low-or-high current division) to painstakingly elaborate (like the list of the IEEE societies). For deciding on a general scope of studies, the partition needs to yield subfields that roughly correspond to the possible courses. The partition used is given in Table 1, and is clearly very informal and imprecise - but formality and preciseness are unnecessary at this point.
Table 1. Subfields and their characteristics
| SUBFIELD | production vs. application oriented | teaching staff availability | market need for engineers | existence of similar courses at the University of Rijeka |
| electric power eng. | application | good | high | the five-semester course at the Faculty of Engineering |
| electric drives eng. | both | good | high | |
| signal processing | production | poor | low | |
| electronics | production | good | average | a short course at the Maritime Faculty |
| control engineering | application | average | average | |
| communications | both | poor | high | |
| high-frequency eng. | production | average | low | |
| computer science | both | average | high | a four-year course combines mathematics and information technologies at the Faculty of Philosophy |
The scope of studies for a university engineering course defines possible positions that can be filled by such engineers, so it has to be made primarily with regard to the industrial environment. The fact that the local electrical industry is application oriented has a profound influence on the choice of the general scope of studies. As can be seen in the Table 1, strictly production oriented subfields had to be discarded.
In addition, the available teaching staff at the University (strengthened by colleagues from adjacent universities) has to be able to implement a course in a chosen subfield. Non-availability of the appropriate teaching staff is prohibitive for a particular subfield (for instance, communications).
Finally, existence of a similar course within the University is a strong reason against a subfield. Computer science is represented in a software-oriented course, and since hardware-oriented computer science is predominantly production oriented, computer science can also be discarded, despite the high market demand for computer engineers.
When the process of discarding the subfields that for various reasons could not (or should not) be implemented is finished, we are left with the following subfields:
and the main scope of studies has to lie within these three subfields.
Strategic design encompasses decisions that influence curriculum in its entirety. Subjects of strategic design can roughly be divided in two categories:
The curriculum-internal strategic design defines the overall structure of university courses in a given field. Its particular importance results from the visibility of its final product - this part of the design defines most of the things that are seen by the students as their "structure of studies".
There are several, mutually dependent issues that have to be resolved:
Greater depth of studies gives future engineers intimate knowledge of a small, specific subfield, and it takes a considerable number of such highly specialized courses to cover the entire required scope of studies. Also, over-specialization is a hindrance when it comes to application in engineering, which usually calls for a broad understanding of subjects. Since this curriculum has to be designed for future application engineers, broadness of knowledge must be pursued.
The duration of higher education studies is usually between two and six years. According to the Croatian law, undergraduate university studies must last at least four years. Current practice in electrical engineering education calls for between four and five years of study, which is in accordance with a recommendation by the National Council of Higher Education that engineering education should not be longer than five years. Since there are two other curricula already established at the Faculty of Engineering, which last nine semesters, organizational reasons suggest that the same duration should be chosen for the new curriculum. Thus, the new curriculum should fit into nine semesters.
The usual duration of courses at higher education institutions is one or two semesters, and sometimes three or four. In our experience, even two-semester courses are too long for engineering education, primarily because they do not correspond to the modern engineering practice, which usually gives an engineer just a few months to master a new subject. Therefore, all of the courses should be designed to fit into a single semester.
Tied to the previous question is the number of courses that a student has to enroll during the course of studies. The decision to keep the courses short increases their total number. A recommendation by the National Council of Higher Education limits the total number of courses for engineering studies at around 45, which gives about five courses per semester.
It would be impossible to cover the required scope of studies in a single course of studies - each of the three chosen subfields can easily constitute a complete course by itself. However, the introductory knowledge is the same, and has a significant overlapping with the existing machine and naval engineering courses. Therefore, a separation point has to be instituted, from which several courses of studies lead to different branches within the chosen part of the electrical engineering. The placement of this separation point has proven to be a difficult decision, and different separation points were attempted during the design. In general, we can have:
Early separation (after the first year of study) was discarded, because the chosen subfields are not independent enough.
Late separation (after the third year of study) was given much thought, and even attempted as a gradual separation during the eight semester in an earlier version of the curriculum [2], primarily because of compatibility with the existing machine engineering curriculum that practices late separation. Unfortunately, the end result showed problems, and it seems that (undergraduate) machine engineering has a much more compact body of knowledge than electrical engineering - the (three) courses of study were separated too late to make a real difference within the electrical engineering.
Halfway separation (after the second year of study) into two courses of study turned out to be appropriate for our intentions. It was decided that both courses should have electrical drives and machines as a common ground, and from there one course would lead towards electrical plant design (by emphasizing power engineering), and the other towards electromechanical system automation (by emphasizing control engineering).
Nearing the end of their studies students should be given the possibility to direct themselves towards their personal professional interests. A balance between compulsory and elective courses should be achieved, to permit specialization without losing the focus of studies. From the initial idea of having three, or even four places reserved for elective courses during the last semester [2], we have settled with the more conservative idea of reserving one or two places for elective courses during each of the last three semesters.
Relationship-based strategic design directly couples the new curriculum with its environment. Main issues are:
Other courses of study offered by the Faculty of Engineering are mechanical engineering and naval architecture.
There is substantial overlapping of introductory courses, mainly mathematics and physics, but also in basic engineering mechanics, computing, project design, foreign languages... Such courses can be reused for the new curriculum, with the added benefit that these courses are already established.
Another possibility for reuse is near the end of the studies, when compulsory courses from other curricula can be used as elective courses in the new curriculum.
Also, in the future it would be possible to use electrical engineering courses as new elective courses for the mechanical engineering and naval architecture studies.
Three models of higher education courses of study can be identified:
Short model is not longer than three years, so it was not an option for this curriculum.
Long-industrial is a course of study designed in tight cooperation with interested industry, in which three to four years of study at the university are followed by a year of industrial work, during which students gain practical knowledge through the mechanism of mentoring, and prepare for their final exam. After graduation, it is plausible that they will find employment in the companies in which they have worked during studies.
The main prerequisite for this kind of industrial cooperation is the existence of strong industrial base with enough potential mentors. Non-existence of such strong base of electrical industry is the reason why this model had to be rejected.
A model in which most of the teaching is performed at the university seems to be the only choice we had for the curriculum. This kind of education gives the teaching staff great control over what will be taught and how, but with a hidden danger of detaching future engineers from the engineering practice - something that can only partly be avoided by careful design of the curriculum. In order to give students the opportunity to see engineering practice before graduation, a special "Engineering Practice" course should be introduced, but outside semesters - perhaps during the summer after the second or the third year of study.
While unable to change the subfields of studies, strategic decisions have profound effect on their representation. Several approaches have been put through succeeding phases of design (in various draft versions of the curriculum), and this point has been shown as pivotal to the entire design, with far reaching consequences. One can hardly overstress the importance of putting much time and effort into strategic decisions, as they essentially define the general framework for a curriculum - a single late change in them is likely to force changes almost everywhere and result in a serious loss of time and effort (as we have learnt after recognizing the strategic fault of late separation in [2]).
Since we have decided on two courses of study, both starting from electrical drives, and then moving towards electrical plant design and electromechanical system automation, we can formalize the decision by naming them:
In terms of strategic design, the framework for our curriculum can be described as:
Tactical design is supposed to implement the decisions made during the strategic design phase. In addition to that, in our experience tactical design often renders strategic design flaws visible, thus serving as a test of strategic concepts.
The body of tactical design is a combinatorial problem of defining particular courses and assigning them to their proper places in the university course's structure, according to the strategic decisions. The complexity of this part of the design lies primarily in its necessarily large group of simultaneous authors, each working on between one and ten courses. The most common problems are:
A strong tool for preventing, recognizing and eventually solving this kind of problems are dependency graphs (Fig 1), showing all allowed dependencies between the courses.
Figure 1. A typical courses dependency graph - dependencies for Computer Architecture
Another kind of problems that can be generated during the tactical design is detachment from the strategic decisions, usually in a form of an unbalanced curriculum. While the former types of problems were localized to a single or just a few courses, this kind is global, which makes it difficult to detect - all the courses have to be taken into account in order to recognize a possible problem. A useful tool in balancing a curriculum is a simple statistical analysis (Table 2) with visualization (Fig 2), in which an unbalanced curriculum would become obvious. For balancing, each course has to be assigned to its class(es), and then the balance of classes is sought according to the strategic decisions.
Table 2: Simple statistical analysis of the curriculum
Electrical Plants:
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 1-8 sem. | % | 5-8 sem. | |
| G | 16 | 11 | 10 | 11 | 2 | 2 | 52 |
23.7 |
4 | |||
| M | 5 | 3 | 3 | 4 | 15 |
6.8 |
4 | |||||
| E | 7 | 7 | 7 | 6 | 27 |
12.3 |
6 | |||||
| EL | 4 | 8 | 8 | 20 |
9.1 |
8 | ||||||
| EP | 7 | 16 | 12 | 35 |
16.0 |
35 | ||||||
| ED | 5 | 5 | 5 | 6 | 21 |
9.6 |
21 | |||||
| C | 3 | 5 | 5 | 4 | 1 | 18 |
8.2 |
10 | ||||
| AC | 5 | 2 | 7 |
3.2 |
2 | |||||||
| X | 2 | 3 | 5 |
2.3 |
3 | |||||||
| el. | 5 | 5 | 9 | 19 |
8.7 |
19 | ||||||
| hours | 26 | 28 | 26 | 27 | 26 | 25 | 26 | 26 | 9 | 219 |
Plant Automation and Mechatronics:
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 1-8 sem. | % | 5-8 sem. | |
| G | 16 | 11 | 10 | 11 | 2 | 2 | 52 |
23.5 |
4 | |||
| M | 5 | 3 | 3 | 11 |
5.0 |
0 | ||||||
| E | 7 | 7 | 7 | 21 |
9.5 |
0 | ||||||
| EL | 4 | 8 | 12 | 24 |
10.9 |
12 | ||||||
| EP | 5 | 2 | 7 |
3.2 |
7 | |||||||
| ED | 5 | 5 | 10 | 20 |
9.0 |
20 | ||||||
| C | 3 | 5 | 3 | 6 | 4 | 6 | 27 |
12.2 |
19 | |||
| AC | 5 | 5 | 5 | 7 | 13 | 35 |
15.8 |
30 | ||||
| X | 2 | 3 | 5 |
2.3 |
3 | |||||||
| el. | 5 | 5 | 9 | 19 |
8.6 |
19 | ||||||
| hours | 26 | 28 | 26 | 27 | 27 | 26 | 26 | 26 | 9 | 221 |
| G - general and introductory subjects | ED - electric drives and machines |
| M - machine engineering | C - computer science |
| E - general electrical engineering | AC - automation and control |
| EL - electronics | X - miscelaneous |
| EP - electric power engineering | el - elective courses |
Figure 2. Graphs of balancing the courses of study
According to a recent recommendation by the National Council of Higher Education, every curriculum must contain a section on necessary, available and optimal resources, and must be accompanied by a separate Curriculum Feasibility Study, before presenting it to the the Council and to the Ministry of Science and Technology with request for starting a new or modified course of study.
In consequence, a complete and detailed plan for the implementation of the new courses of study had to be made [6].
Resources can be divided into:
Human resources are the most critical part of any education, one that cannot be improvised nor substituted. As noted in section 3, availability of proper teaching staff has been taken into account since the beginning of the curriculum design, and only those subfields that had at least an average staff availability could pass to subsequent phases of the design.
Introductory courses are well covered (90% coverage in the first four semesters with teachers already employed at our Faculty). Special problem is posed by the desired broadness of studies, which calls for many teachers from different fields. In order to facilitate this broadness, a total of around 40 professors and 25 assistants will be needed (the numbers were obtained by counting the names that stand next to the planned 79 courses). Total course coverage with the teachers from our Faculty (for both courses of study and all nine semesters) on October 1, when the studies are supposed to begin, is estimated at around 70%, and the rest will have to be organized in due time through inter-university cooperation and affiliates from industry.
Space has to be made available for lectures, laboratories and additional staff, and the way to ensure the space requirements is to include a syllabus with a space plan as a supplementary part of the curriculum.
Since the studies are supposed to be university based, they will depend on good laboratories. In a fast changing field like electrical engineering, even a teaching laboratory can never be finished, because it has to change together with the engineering practice. Because of that, it is important to start organizing any particular laboratory as late as possible (to minimize losses from equipment getting out of date), and yet early enough to have a functioning laboratory when needed for its courses. From experience gained so far, we estimate that the optimal time to begin organizing a laboratory is about a year before it is needed, and we plan a total of seven student laboratories (existing two, and five to be furnished during the next three years).
Financial resources can be further divided into three parts:
One additional resource that never stands by itself, but always coexists with other resources is time. All other resources require substantial investment of time, and proper timing is supposed to be essential once the courses actually begin. Unfortunately, there are too many unknown parameters (probable financial problems, private problems of the essential staff, other obligations, equipment problems or unavailability... to name just a few), and time cannot be really planned within the curriculum or any accompanying document.
A design process is really the first thing that has to be designed when preparing the writing of a curriculum. Very few documents (mainly in the area of standardization) call for such a diverse and extensive group of authors, where each author's work influences a significant part of the document, sometimes in an indirect and concealed way.
To put the problem into perspective, the list of people who worked on the curriculum comprises:
Also, for administrative reasons, each revision had a hard deadline for submission to the University Senate and the National Council for Higher Education. Therefore, semiformal design process of the curriculum and accompanying documents had to be instituted.
Figure 3. Dependency graph for parts of the design process and documents
The design process is highly iterative. Due to the fact that iterations are made in a distributed environment of at least five, and sometimes as much as forty authors, a mechanism for enforcing, checking and reestablishing conformance with basic ideas, as well as a mechanism for finding internal incongruities have to be instituted.
One important tool is a dependency graph showing parts of documents and design process together with ways through which they can directly change one another (Fig. 3). Changes can then be traced (both forward and backward) along the dependencies.
Another important part of the process is issuing minor revisions. A minor revision is issued for internal reviewing only, but in a form as close to the final curriculum as possible, and represents a checkpoint for the completeness of the work. With a rate of one minor revision issued every three or four weeks we typically had two or three minor revisions for each major revision, making the design process traceable and predictable, and thus facilitating conformance with the deadlines.
A complete trace of changes, faults, suggestions, questions and decisions was kept, and most of it could be (and often was) traced to their origin.
Design process of an electrical engineering curriculum doesn't end with course inception, because the field itself is changing rapidly. New iterations will certainly be needed in the course of the coming years, to keep our university course from getting obsolete.
The new curriculum has been designed, comprising two electrical engineering courses of study, in a reasonably orderly and predictable manner, and each revision was delivered on time. Decision making process was well documented, and all important decisions are traceable. Due to semiformal design process, it was possible to rebuild major revisions in relatively short (sometimes at the order of a single month) and, even more important, predictable time, giving what we believe is a high quality final product.
While final conclusions can only be given after several generations of engineers graduate the new university courses, we believe that semiformalism and iterative development added quality to our design, primarily through better project management and concentration of effort on identifiable parts of the complexity of creating a completely new curriculum.
The course that would be the implementation and, at the same time, the true test of our design is expected to begin this autumn.