System Design and Complexity

Setting the objective, search for solution, selection of solutions The steps are prerequisites of the following step, proper preparation helps (Roli: maybe it not only helps but is crucial?) for following tasks

Inner system: part of the system which is affected by the problem and its possible solutions. Environment: contains system. Inner system can be enlarged (size can be corrected due to the effect of the solution)

Solution neutral, describe requirements, operable, measureable, complete, no contradictions, no redundancy, prioritized.

Types: procedural, system, mandatory, desired

As a help to formulate objectives (compatibility, dimensions of objectives)

1. Order: modification within system, 2. Order: modification of system rules

Synthesis: generate solutions, analysis: evaluate solutions, Create variety vs. reduce variety, Iterative process

Need to know time needed for an activity, cope with uncertainty of activity duration, dependencies between activities

Relationships between activities are not shown in Gantt, impact of delay is not shown, CPM shows critical path, allows capacity planning

Longest path through schedule, no float

Time by which an activity can be postponed, float gives flexibility in scheduling activities

Point for go/no go decision; because of previous investments -> resources to complete the project decrease

Planning: go back to earlier stage, controlling: only affect implementation, doesn’t require to go back to planning

Small changes are amplified due to feedback loops. By trying to balance system unintended, destabilizing effect/loops emerged

Constraints and feedbacks between upstream and downstream project phases -> concurrent processes

Internal: within a phase -> infeasible area

External: between different phases -> not completely dependent

Due to complexity as a system property

• positive feedback loops => instability (undiscovered rework)

• Small deviations can get amplified

• Negative feedback loops induce additional positive feedback loops (overwork => burnout)

• Time boundedness: role of correlations and delays (downstream processes feedback to upstream processes).

Micro: large number of strongly interacting elements, interactions of the elements

Macro: dynamics of the system as a whole, dynamics and properties of the whole system, and focus on a few representative agents

System dynamics => Macro-Level

Positive causation: enhancing effect

Negative causation: diminishing effect

Quantification: derivative of function

Increase/decrease of two (independent) variables is correlation, the same (identical) causes lead to the same (identical) effects is causation

Positive feedback: enhancing/reinforcing

Negative feedback: diminishing/balancing

Positive: population growth → growth of birth rate

Negative: death/consumption/overcrowding

(i) Decoupled:

Rabbits: birth vs. death rate (=> birth > death rate => exp. growth) and carrying capacity (=> logistic growth, s-curve).

Coupled: Rabbit pop depends on fox pop and vice versa. Oscillations depend mostly on rabbit birth rate

3 cases:

strong changes but still periodic

regular oscillations

stationary solution (dying out or decoupled system with carrying capacity).

Target inventory adds an additional time scale → oscillations

 Stock is not simply used, but build up of inventory leads to oscillations

When increasing sales lead to an adjustment in inventory → needs an overshoot in production

Slow-moving: sales fall short of forecast, lag of production planning

Hot selling: too large lead time leads to phantom ordering (and order cancellation)

Balance with inventory correction

Stand-alone: look at measures separately

Integrated: model analysis to profit from synergies

Solving bottlenecks can create new ones

Introduction, growth, maturity, saturation,

Introduction, growth, maturity → question mark, star, cash cow

Introduction, saturation → question mark, dog

the different life cycles have all the bell innovation curve in common, with an initial phase of early adopters, followed by growth phase and the maturity phase and finally faded out in the saturation phase.
the overall demand life cycle of a dedicated technology (e.g. audio technology) can be extended with the combination of new technology and product life cycles that extend a preceding one and therefore lead to an extended demand curve.

The bass innovation model assumes that there is a constant population comprised of two groups adopters and potential adopters. Every potential adopter becomes a adopter over time. The number of potential adopters can only decline and the one of adopters only increase.
No discarders are taken into account, no re-buyers are neither taken into account...

The technology diffusion considers the spatial distribution of a product, how the technology diffuses in space.

• bass innovation model has a cold start, as it starts with one adopter and slowly rises. The higher the contact rate and the probability to buy the product, the steeper the slope becomes.

• the common source model uses the broadcasting effect (e.g. by advertising a product) and therefore does not have a cold start but rather stars growing very steep. There is no s-curve possible with this model

• the mixed source model is a combination of the bass innovation model and the common source model and depending on the parametrization either of the effect is stronger.

• the density dependent model takes a ceiling, the carrying capacity into account as well as a birth/death rate which determines the slope of the curve.

• The technology diffusion model does not take new arriving technologies into account

• It is assumed that new technology will not fail

• Compared to the symmetric curve in the technology diffusion, the empirical model shows asymmetric s-curve

• competition between technologies is not considered

Bifurcation: branching from one stable, to two stable solutions; two parameters → Asymmetric potential

The logistic map shows discrete dynamics. The control parameter influences the max/min of the plot, though not its general shape, it also determines the time scale.

For larger r additional solutions can be found ( → oscillations)