Contact Person
Dipl.-Inf. Michael Wagner
Senior Scientist
Business Unit SQC
+49 30 3463-7391

Funded by
BMBF Logo Artemis Logo

VARIES

VARiability In safety critical Embedded Systems

May 01, 2012 to Apr. 30, 2015

Profitability through optimal variant diversity of safety critical embedded systems

Embedded systems have become omnipresent components of various products. They perform in TVs, mobile devices and washing machines, but also in safety critical products such as Driver Assistance Systems (airbag control, ESP, ABS) or medical devices (infusion pumps, heart-lung and dialysis machines). Particularly in the latter, human life may depend on reliable performance of the systems. Therefore, they underlie various regulations and certification requirements, where the in these fields of application required functionality and reliability must be verified. At the same time, the producers of such systems face the challenge to provide their innovative and market-driven products of high quality in a time and cost efficient manner. Thus, the embedded systems used in these products usually are not developed from the ground up, but reuse approved solution components.

By the introduction of innovative software engineering approaches in non-safety critical fields (e.g. introduction of product lines, configuration management, modular systems) the development effort could be more than halved. Within the VARIES project 23 partners from seven European countries – including five partners from Germany – work on maximizing the potential of variability in safety critical embedded systems, which underlie additional requirements.

To reduce the effort of certification of safety critical end-products while having high variability and variety of products, adapted and expanded software technologies and methodologies are required. Therefore, the VARIES project team will develop an extensible reference platform for the "Variability Management". Various technologies for the development of product versions and the reuse of software components of different domains can be integrated via this platform. So, the information exchange between the various components will be enabled. Furthermore, VARIES aims to establish a „Center of Innovation Excellence“, where the concentrated knowledge about reuse and variant management is provided for interested parties from industry and research.

Partners:

  • pure-systems GmbH
  • Berner & Mattner Systemtechnik GmbH
  • TÜV Süd AG
  • Atego Systems GmbH

Innovation

The main innovations introduced by VARIES will be:

  • The VARIES project has the ambition to deliver a complete reference platform for managing variability in safety-critical embedded systems. Special attention will be given to the creation of safety-critical systems through a Product Line approach.
  • VARIES will achieve improved interoperability and industrial impact by building on already existing standardization efforts in the area of product lines.
  • In order to succeed or even to survive, manufacturers and system integrators must be able to deliver new products with speed, diversity, high quality, and at an acceptable cost. The VARIES project envisions to manage product & processes with an increase complexity with reduced effort figures, dealing with uncertainty while maintaining an independent hardware and software upgradability all along the life cycle.

Expected Results

The main results of VARIES will be that

  • Innovative embedded systems (ES) will develop a reference framework for lean innovation, where the framework is made of new models, tools, and design methods,
  • The VARIES Platform: a complete, cross-domain, multi-concern, state-of-the-art reference platform for managing variability in safety critical ES will be delivered. Special attention will be given to aspects specific to safety critical ES, in particular the impact of reuse and composition on certification. In this platform different tools can be instantiated and chained to support the process flow of development for embedded systems product variants, over the whole product lifecycle, tailored to the specific context to a given company and
  • A Centre of Innovation Excellence on the topic of how to come to an optimized product variability that can be used for a variability of markets will be established.

Results

Framework For Identifying And Describing Resources And Costs Across The Product Lifecycle

1 Framework for identifying and describing resources and costs across the product lifecycle

The text below has been reproduced from VARIES deliverable D3.6 [Biot, 2015], section 8, and has been adapted for public dissemination.

1.1 Introduction

When assessing the prospect of increasing variability for any product in any organization, it is important to look at the costs. As the variability paradox states, introducing variability should provide benefits, but also increases costs, see deliverable D3.1 [Hirvonen, 2013]. The relation between costs and variability also seems to be non-linear – if variability keeps increasing, complexity and the ensuing cost increases grow (see e.g. Strategies). When the company feels pressure to decrease costs, this cost pressure in itself becomes a variability driver (e.g. drivers companies to decrease variability).

The effect of complexity can also increase costs in unpredictable ways, create bottlenecks in critical resources and raise the base costs for production. This makes the cost driver a complex and important one, and it raises questions to be assessed; how much will it cost to increase variability in the product portfolio, and what do these costs consist of? Which types of costs are the most sensitive to variability and how do these costs formulate? What will be the probable resource bottlenecks, and how will these bottlenecks affect the time cost for operations? How will increasing variability in the product portfolio affect the resource needs and thus costs in time and money later in the lifecycle?

The current framework is an attempt to analyse the effects of variability on resource consumption when creating embedded systems in different life cycle phases independent of domain. This framework assesses the cumulative resource cost of variability in an organization that develops, produces and performs maintenance of products. The approach of the framework is identifying the types of resources, and the phases and tasks in the life cycle of a variant, that are most sensitive to the effects of variability.

It should be noted, that this approach observes resource cost effects of variability, not all costs or their effects on products. The reason for this is simply the complexity of handling all possible costs in all possible domains. Due to this approach this section is about the resource costs of variability and handling of those costs.

1.2 Approach

All activities in a company consume resources. These resources include time of employees, used facilities and materials, used services, subcontractors etc. Some of these resources generate costs directly when consumed, while some of them just replace the availability from other activities. In this framework we are interested in resources that are somehow related to variability of products. Thus, different types of resources and resource costs related to products are of interest in this section.

Normally a company has several different product variants that are in different stages of their life cycle. While some of these products or their variants are becoming obsolete, other variants might be in the maintenance-, production- or development- stage.

During different stages of a product life cycle different resource costs emerge and dominate the accrued resource costs of a product or a product variant. To understand which resource costs are most significant in which life cycle stage of a given product – resource costs in different stages of the product life cycle (for different kinds of products and product variants) need to be examined separately.

In the framework we (1) assess the resource costs of variability in different life cycle stages and the tasks included in those phases, and (2) assess which of these resources costs will potentially create bottle necks or snow ball effects when the level of variability changes, and (3) try to find the possible trade-offs for organizations to tackle the resource bottlenecks.

The life cycle model used is based on Blanchard [Blanchard, 2010]. The full life cycle with adjustments introduced in deliverable D3.1 [Hirvonen, 2013] includes the categories shown in Figure 1‑1 below.

Figure 1 1. The full life cycle costing categories by Blanchard with changes introduced in deliverable D3.1 [Hirvonen, 2013]; see also deliverable D3.6 [Biot, 2015], section 5.3

1.3 Framework for Analysis

In this subsection different approaches for analysis are explained, as well as the relevant cost types that are important in variability resource cost analysis.

1.3.1 APPROACHES FOR ANALYSIS

The approach to identify the resource needs and associated costs impacted by changes in variability. These resource needs are categorized according to the variant life cycle and according to the tasks in that life cycle, both based on Blanchard [Blanchard, 2010]. The idea is to identify key resource costs that are impacted by adding new product variants, modifying existing variants, or eliminating variants. Several different resource costs are always affected, but the idea is to assess whether these resource costs create a choke point, cause delays, or create a snow ball effect to any resources or costs. These resource costs are the key resource costs.

To achieve this analysis, several known cost effects of variability are applied to the lifecycle. In the following these costs are explained.

1.3.2 FLEXIBILITY COSTS

According to Pohl [Pohl, 2005] there are costs caused by a high quantity of variants for different departments of a company. These are costs that typically increase as the number of variants increases. These kinds of costs are called flexibility costs and are described to accumulate for different departments of a company.

1.3.3 OPPORTUNITY COSTS

Another important, if more generic, cost type is opportunity costs that occur always when one variant is developed, produced or maintained in place of another variant.

The notion of opportunity cost plays a crucial part in ensuring that scarce resources are used efficiently. Thus, opportunity costs are not restricted to monetary or financial costs: the real cost of output forgone, lost time, or any other benefit that provides utility, should also be considered as opportunity costs [Green, 1894].

1.3.4 DISTRIBUTION COSTS

According to de Groote and Yücesan product variety increases logistics costs [deGroote, 2011]. When dealing with a higher product variety, the level of inventories and lead time is increasing. These respectively mean that the logistics costs and logistics time to deliver products increases. A company could choose to keep inventory costs low and not care about increases lead times, which would however lead to lower service levels and dissatisfied customers.

1.4 Resource cost impact of variability decisions types

To assess the resource costs as impacts of variability decisions, the impact of different variability decision types is analysed for each costing category in the model. The purpose of this observation is to emphasize and to clarify the relationship between the variability decisions and the resource costs.

Linking all of the costing categories or tasks to all of the variability decision types would make this analysis massive and messy. Thus a more subtle and effective approach is needed. To do this, the two-dimensional classification of product differentiation versus product variety is utilized (D3.6, section 4.1, Figure 4 1 [Biot, 2015]).

1.5 trade-offs

There are several resource cost affecting impacts of variability decisions, which often entail a trade-off. Because the framework analyses resource costs and their impact, also these trade-offs need to be analysed, as many of them are done already when variability decisions are made and the resources are consumed. There are also several different ways in which variability decisions can impact a company (e.g. cost, revenue, time-to-market …) (see e.g. D3.6 chapter 4 and section 9.5 [Biot, 2015]). In addition, variability decisions often have both positive and negative consequences (i.e. there is often a trade-off between different decision consequences). Derived from these impacts, a group of trade-offs are presented in the framework and applied to the framework. These trade-offs include:

  • Resource capabilities (manpower/skills/production equipment) versus investment/cost
  • Variability to gain market share/revenue versus costs
  • Commonality versus diversification: platform versus customisation cost
  • Commonality/cost versus quality/product performance

1.6 Application

Behaviour of different costs are analysed per task in a life cycle phase according to variability decisions in the framework, and trade-offs are discussed. This framework of resource cost behaviour then creates a base for determining resource cost impact for any company on a rough level.

2 References

Biot, 2015 Biot, O., Bollen, M., Codenie, W., Hirvonen, H., Teppola, S., Van den Broeke, M., Vierimaa, M. (2015). Deliverable D3.6: Updated methods to capture product variability: drivers and variants.

Blanchard, 2010 Blanchard, B. S., & Fabrycky, W. J. (2010). Systems Engineering and Analysis. Prentice Hall.

Green, 1894 Green, D. I. (1894). Pain-Cost and Opportunity-Cost. The Quarterly Journal of Economics, Vol. 8, No. 2 (Jan., 1894), 218-229.

Hirvonen, 2013 Hirvonen, H., Mäntysaari, M., Pesonen, K., Haapaniemi, A., Biot, O., González-Deleito, N., . . . Jaring, P. (2013). Deliverable D3.1: Modeling and evaluation of variants in the product portfolio and roadmap

deGroote, 2011 de Groote, X., & Yücesan, E. (2011). The impact of product variety on logistics performance. Proceedings of the 2011 Winter Simulation Conference (pp. 2245-2254). IEEE Conference Publications.

Pohl, 2005 Pohl, K., G. Böckle, and F. van der Linden. 2005. Software Product Line Engineering: Foundations, Principles, and Techniques. Springer.

VARIES-TA ARTEMIS Call 2011, ARTEMIS-2011-1, 295397 VARIES: VARiability In safety-critical Embedded Systems. Technical Annex.

High-Level Strategies for Addressing the Variability Paradox

1 High-Level Strategies for Addressing the Variability Paradox

The text below has been reproduced from VARIES deliverable [D3.1], section 2.3, and has been adapted for public dissemination.

1.1 Product Variability

Product variability is defined as the art of managing the creation, evolution and maintenance of different ‘versions’ (variants) of a product [VARIES-TA]. This definition encompasses the creation of new products in the portfolio, the sequential incremental updates of a product (i.e., evolution) as well as the creation of derived products (i.e., customization and configuration). We follow the definitions given in [Codenie, 2009]: “the term customization is used to refer to the activity of changing a (generic) product into a solution satisfying the specific needs of a customer. Changing could be adding new functionality, and changing or removing existing functionality. In essence, customization creates a new product variant that not existed before. Configuration is choosing among a predefined set of product variants (e.g. by filling in configuration parameters).”

1.2 The Variability Paradox

Figure 1 below illustrates the typical course of value (turnover, profit) and cost curves related to the product variety:

  • A number of reasons for creating product variants are market segmentation and keeping up with evolving market trends and requirements. Typically, the turnover will increase by creating variants tailored to these new market segments or that provide an answer to upcoming market needs. However, adding more variety will eventually provide fewer yields: the turnover curve (green) evolves almost logarithmically as variety increases [Schuh, 2005]. This is consistent with the typical market segmentation story.
  • By adding more variety, initially the engineering of the first variants can benefit significantly from acquired product expertise, domain knowledge and reuse & adaptation of existing engineering artefacts. However, by adding more variety in the portfolio, the incremental cost for creating one extra variant starts growing exponentially (red curve), since product engineering will have to manage the overhead caused by interdependencies between different product variants. This is exacerbated when dealing with multidisciplinary and multi-technology products having a life expectancy exceeding the lifecycles of the technologies used.
  • As a result, the incremental profit (margin) curve (blue) will show a maximum and will gradually evolve to a point where added variety will have negative yield (red zone).

On this graph there appears to be an optimum (in the green zone) where variety contributes most to the yield. To the left of this optimum, yield can effectively be increased by adding more variety (e.g., via market segmentation). To the right of this optimum, yield can still increase by adding more variety, however the extra variants will contribute less and less to the overall profit. Eventually the yield of extra variety will no longer outweigh the penalty of the increased complexity of managing the extra variability (red zone).

The graph in Figure 1 leads us to the “variability paradox”:

How can the benefits offered by introducing variability into embedded systems

outweigh the increased product complexity caused by variability?

A first challenge is to try identifying the optimal level of variety. Mathematically it’s where the blue margin curve’s first derivative is zero. However, in practice, this curve depends on many more, often intricately interdependent parameters than variety and cost alone, hence finding this optimum is difficult.

A second challenge is that the situation evolves over time. For instance, a once profitable market segmentation and assorted offering may no longer be profitable because of competition leading to cheaper products that lower the green curve and consequently reduce the margin, sometimes even to the point of generating losses (see Figure 2 below).

This reasoning is well described in literature.

1.3 Strategies for Addressing the Variability Paradox

Based on Figure 1, a number of high level strategies can be proposed to address the ‘variability paradox’:

  • Strategies that do not affect the cost and value curves
    • (1.3.1) Add more variants
    • (1.3.2) Eliminate unprofitable variants
  • Strategies that affect the cost curve
    • (1.3.3) Reduce costs of variability
    • (1.3.4) Manage variability
  • Strategies that affect the value curve
    • (1.3.5) Make better products
    • 1.3.1 ADD MORE VARIANTS

A first strategy is to add more variety in the product portfolio. This can be applied in cases where the benefits of the extra product variety outweigh the engineering & management effort for making and maintaining these new variants. A typical scenario is the identification of a significant market segment in an already served market for which the current product offering does not satisfactorily meet the needs. In another scenario a promising market niche has been identified, for which it is understood how to adapt existing products and engineering artefacts to build products adapted to that segment’s needs.

To assess the feasibility and yield of this strategy, often a business case is developed.

Figure 3 illustrates how this strategy moves along the existing curves in the direction of increased variability, with the intention to remain in the green area of positive yield of variability.

Figure 3. Strategy 1: add more variants

The following examples illustrate how this strategy may be implemented (this is context dependent):

  • Apply market segmentation. By identifying criteria for segmenting your market, you may gain insight in potential variants that better support a segment’s particular weighed set of needs (certain features may be more or less relevant for different segments).
  • Identify customers with similar needs.

1.3.2 ELIMINATE UNPROFITABLE VARIANTS

In some cases, there is too much diversity in the product portfolio. Some variants cost more to develop and maintain than they contribute to the profit margin. Several scenarios can be proposed in which this case applies. A product variant developed for a specific market segment in the past may incur high maintenance costs due to technology obsolescence; in addition the profit margin may have been put under pressure by competitors. In other cases there may be a motivation to become a dominant player by offering a product for several existing market segments; where initially this made sense and did realize added profit, eventually the market needs changed, competition may have offered a hard to beat alternative and the added complexity of managing all existing variants is further eroding the profit margins.

A first strategy to apply in these scenarios is pruning the product portfolio from the less profitable or loss inducing variants. In Figure 4 this is achieved by moving leftward on the profit curve towards less variety.

Figure 4. Strategy 2: eliminate unprofitable variants

The following examples illustrate how this strategy may be implemented (this is context dependent):

  • Identify legacy products that tend to waste resources in engineering and production merely to support aging, low margin and low volume products. These resources could be reassigned to work on the future products in the portfolio.
  • Make or refine the business plan for problematic products
  • Remove product features that are no longer relevant. This can happen for several reasons, e.g. due to technology obsolescence, new market trends, removing product features that are no longer in use, no longer supported or no longer relevant.
  • Re-segment the market. The criteria used in market segmentation may have evolved to the point where they became irrelevant. This can affect the number of variants to offer.
  • Evaluate how the product variants are actually used

1.3.3 REDUCE COSTS OF VARIABILITY

Instead of walking the existing curves, one can also attempt at manipulating the curves. This leads to the following set of strategies.

Figure 5 illustrates a first strategy in which the cost curve is pushed downward. The goal of this strategy is to produce the same variety at a lower cost. The net effect of pursuing this strategy is to increase the profit margin. Indirectly the green profit zone also grows to the right.

Figure 5. Strategy 3: reduce costs of variability

The following examples illustrate how this strategy may be implemented (this is context dependent):

Deploy configuration management in the product development organization

Identification and management of reusable assets

Optimize the economic return of reuse versus customer specific customization. Reuse comes with a cost, it is important to evaluate whether the reuse will ever recoup this cost.

Deploy instruments and processes to better support the management of the variants

Deploy a test automation framework

Reduction of costs of variants may occur in several ways [Pohl, 2005]:

Upgrading products by standard integration of formerly supplementary features. Reducing complexity, which is caused by supplementary features, reduces production costs.

Modular structure. A modular structure allows high diversity of variants under the condition of low increase of complexity.

Product platforms. The usage of platforms and non-variable parts lowers the diversity of parts and the production complexity.

1.3.4 MANAGE VARIABILITY

The next strategy consists of offering more variety without incurring any cost penalty. Figure 6 illustrates this strategy as the red cost curve being pushed rightward. The net effect is to grow the green profit zone rightward too. Indirectly, the profit curve will grow thanks to the variability management.

Figure 6. Strategy 4: manage variability

Figure 6. Strategy 4: manage variability

The following examples illustrate how this strategy may be implemented (this is context dependent):

Deploy configuration management in the product development organization

Use technologies that support product configuration

Deploy instruments and processes to better support the management of the variants

Adding a layer of abstraction (divide and conquer)

Optimize the definition of subsystem boundaries (e.g., to facilitate up-scaling of the engineering activities for creating more variants)

1.3.5 MAKE BETTER PRODUCTS

Lastly, the turnover curve can be pushed upward by creating products that bring more value to the customers. As a consequence, such products can be sold at a higher price. A direct consequence is that the profit curve will grow upward (Figure 7). Indirectly the green profit zone will expand rightward thanks to the turnover contributing more substantially to the profit margin.

Figure 7. Strategy 5: make better products

Figure 7. Strategy 5: make better products

The following examples illustrate how this strategy may be implemented (this is context dependent):

Build products that better meet the customer needs and wishes. If your offering more closely matches the customer’s needs than other offerings, yours brings more value; hence you may decide to ask a higher price for your offering.

Create products that better support the users in their tasks and activities. Again, these products may offer more value to their users, hence may be a reason for asking a higher price.

Value innovation. In this approach [Kim, 2005], you try to define a unique value proposition based on an existing offering in an existing market. By identifying value elements that differentiate your offer from competing offerings, you can bring uniqueness in your value proposition.

Market segmentation. In some cases, the same product can be sold at a higher price to a specific market segment (e.g. vanity products, ODM/OEM rebranding).

Identification of new market niches. By exploring the relevance of your offering in related and distant market niches, you may end up offering high value by building a product that is targeted at that niche.

Aforementioned high-level strategies serve different purposes and are sufficiently generic to be broadly applicable. The exercise of choosing which strategy to apply heavily depends on the context of the company, which is impacted by the variability drivers at play.

2 References

Author, Year Authors; Title; Publication data (document reference)

Codenie, 2009 Codenie, W., González-Deleito, N., Deleu, J., Blagojević, V., Kuvaja, P., Similä, J. (2009). Managing Flexibility and Variability: a Road to Competitive Advantage. In Applied Software Product Line Engineering, Chapter 12, pages 269–313. Taylor and Francis, December 2009

Kim, 2005 W. Chan Kim, Renee Mauborgne. Blue Ocean Strategy: How to Create Uncontested Market Space and Make Competition Irrelevant. Harvard Business Press, 01 Feb 2005

Pohl, 2005 Pohl, K., G. Böckle, and F. van der Linden. 2005. Software Product Line Engineering: Foundations, Principles, and Techniques. Springer.

Schuh, 2005 Schuh, G., Schwenck, U. (2005). Produktkomplexität Managen: Strategien – Methoden – Tools, Carl Hanser Verlag Fachbuch, ISBN 3-446-40043-5.

VARIES-TA ARTEMIS Call 2011, ARTEMIS-2011-1, 295397 VARIES: VARiability In safety-critical Embedded Systems. Technical Annex.

D3.1 ARTEMIS Call 2011, ARTEMIS-2011-1, 295397 VARIES: VARiability In safety-critical Embedded Systems. Deliverable D3.1: Modeling and evaluation of variants in the product portfolio and roadmap.