Model-Based Design (MBD) and Model Driven Development (MDD)
Comparing Modeling Design Outcomes with Comparable non-modeling Design Outcomes

In a soon to be published EMF white paper, cadres of comparable design outcomes were developed between developers that used MBD tools and those that didn’t. Cadres were established worldwide, for North America, for Asia, and for Europe.
In addition, specific analyses were conducted for MBD and non-MBD cadres for Telecom/Datacom, Medical, Automotive Transportation and Industrial Automation application markets.
Total Cost of Development calculations were conducted using:
- Number of software and hardware engineers per project
- Time from design start to product shipment
- Percent of designs cancelled and the number of months elapsed before cancellation
- Percent of designs completed behind schedule and number of months behind
The following table summarizes EMF’s findings. For comparison, the cost per developer man month was chosen to be $10,000. Clearly, this dollar value is high for Asian developers. However this value was chosen for internal geographic analysis only – to establish whether MBD provided an advantage or not. These values are not to be used to compare, for example, Asian costs with European costs.
|
|
|
MBD |
||
|
Non-MBD Cost |
MBD Cost |
Advantage |
||
| North America |
$3,921,519 |
$3,153,452 |
24.4% |
|
| Europe |
$3,744,894 |
$2,722,134 |
37.6% |
|
| Asia |
$10,189,266 |
$3,374,067 |
202.0% |
|
Looking at worldwide developments (that is interrogating the entire database irrespective of geographic considerations), the following table summarizes the EMF findings.
|
World |
World Industry |
|
|
Industry MBD |
Not MBD |
|
| Devel time Months |
13.4 |
13.1 |
| % behind schedule |
44.3% |
49.9% |
| Months behind |
3.7 |
3.7 |
| Ave Delay Months |
1.63 |
1.83 |
| % cancelled |
10.1% |
12.1% |
| Months lost to cancellation |
4.2 |
4.7 |
| SW Developers/proj |
11.0 |
16.3 |
| HW Developers/proj |
8.9 |
10.9 |
| Total project developers |
19.9 |
27.2 |
| Average Developer months/project |
266.2 |
358.1 |
| Developer months lost to schedule |
32.4 |
49.8 |
| Developer months lost to cancellation |
8.5 |
15.5 |
| Total developer months/ project |
307.1 |
423.4 |
| At $10,000/developer month | ||
| Average developer cost/project |
$2,662,098 |
$3,580,843 |
| Average cost to delay |
$323,977 |
$497,835 |
| Total developer cost/project |
$2,986,075 |
$4,078,677 |
|
MBD Adv |
36.6% |
|
|
|
|
It is interesting to note that in every analysis, regardless of the cadres used (i.e., each vertical or geographic comparative breakout), MBD projects used fewer developers. When analyzing cost overruns (i.e., the number of developer months lost to cancellation or late completion) to total project developer months, the percent of cost overruns to total project developer months was less for MBD in every analysis.
EMF suggests that this data shows that the advantages in using simulation-modeling as a design methodology are real and that these practices will be adopted for reasons not only related to design outcomes, but for financial ROI reasons as well.
Most software is nearly -impossible- to test under flawless conditions. Especially embedded systemused with small amounts of CPU power and memory.