SoftwareReleaseProject

Causal loop example given by van Zijderveld (MARVEL)

Diagram

Information

This information is part of the Business Simulation Library (BSL). Please support this work and ► donate.

In 2007 Erik J.A. van Zijderveld introduced what he coined a "Method to Analyse Relations between Variables using Enriched Loops (MARVEL)" [24]. He made the following observations:

A causal loop diagram (CLD)—or impact diagram—is a good tool for integrating information from different stakeholdes and a modeling paradigm, which people with different backgrounds can easily relate to.

While quantitative data may be hard to obtain quickly and reliably, stakeholders and decision makers can often reach consensus on speed and strength with regard to impact relations.

Decision makers can typically identify pressure or intervention points and relevant performance indicators in such diagrams.

Often to come up with relevant insight it will suffice to show initial dynamic behavior, i.e., we do not need the kind of long-run validity that a fully fledged simulation model provides—given that the effort building such a model will usually be 10 to 100 times higher.

In section 5 of his paper van Zijderveld gives an illustrative example for the application and utility of MARVEL. The MARVEL diagram for that illustrative application is given below. The problem at hand is a software release project, i.e., a new software is to be introduced in a company, which has great impact upon procedures and organization, while resistance against the software endangers the project. The resistance stems from dismissals, low software quality and poor management [24, p.13].

MARVEL Diagram for a new software release project [24, p.14]

In the diagram, the thickness of an arrow indicates the strengh of impact whereas length and number of "barriers" at their tips indicate the speed of impact propagation. These attributes are assigned qualitatively using weak, average, strong, very strong to rank strengths and low, average, high, very high to rank speeds.

The model diagram below illustrates how such a model can be built using the →CausalLoop package. Impact relations between variables are encoded using the →ProportionalityDelayed component with global parameters for strength (wk, av, st, vs) and speed (v1, v2, v3, v4) in place, the latter being transformed to delay times, which are inversely related to speed.

Diagram View

In accordance with the MARVEL version there are four control or intervention points, which can be turned on or off for a simulation run:

c1 : budget control with setpoint = 0.8

c2 : mandate control with setpoint = 0.7

c3 : management knowledge level control with setpoint = 0.5

c4 : management quality control with setpoint = 0.6

These controls mean that whenever there is a deviation between setpoint and actual level of the variable under control, then corrective action will be taken, i.e. a flow to the stock, in order to erradicate the deviation within the chosen adjustment time, whcih is set to 1 yr for all controls.

There are three main performance goals: cost effectiveness (costs), production quality (quality), and software usage (usage). In this example the normalized stock values pertaining to the goals are simply mapped upon the linear scale [0,1]. The graph blow shows the development for these performance indicators over a period of 10 years with just the budget control (c1) activated.

Performance Evaluation (c1 active)

In order to better combare different scenarios we can use a weighted average performance and calculate the mean over the simulation period (→totalPerformance). Using this measure we can compare different combinations of interventions and it turns out that a combination of c3 and c4 shows best average performance albeit withouth considering the control effort (see Notes).

Performance (basis points) for different control combinations

Notes

The parameters in the model are set to the values specified in van Zijderveld's paper whenever possible.

The model by specification is set up in equilibrium as there are not processes of exogenous growth in the model and all motion is triggered by changes. If at least one control is activated the model will show dynamic behavior.

Since control elements (optionally) report their rates, i.e., the control effort, one can readily expand the evaluation of scenarios by comparing performance gains to the required control effort.

Parameters (15)

wCost	Value: 1 Type: Real Description: Weigth for cost performance
wQuality	Value: 1 Type: Real Description: Weigth for quality performance
wUsage	Value: 1 Type: Real Description: Weigth for usage performance
wk	Value: 0.2 Type: Real Description: Weak influence
av	Value: 0.5 Type: Real Description: Average influence
st	Value: 1 Type: Real Description: Strong influence
vs	Value: 1.2 Type: Real Description: Very strong influence
v4	Value: 7889238 Type: Time (s) Description: Very high speed
v3	Value: 15778476 Type: Time (s) Description: High speed
v2	Value: 31556952 Type: Time (s) Description: Average speed
v1	Value: 94670856 Type: Time (s) Description: Low speed
c1ActiveQ	Value: true Type: Boolean Description: = true, to activate the budget control
c2ActiveQ	Value: false Type: Boolean Description: = true, to activate the clear mandate control
c3ActiveQ	Value: false Type: Boolean Description: = true, to activate the management knowledge level control
c4ActiveQ	Value: false Type: Boolean Description: = true, to activate the management quality control

Connectors (1)

	modelOutput	Type: DataOutPort Description: The model's performance output

modelOutput

Type: DataOutPort

Description: The model's performance output

Components (73)

	modelSettings	Type: ModelSettings Description: Setting important global variables within a model
	lab1	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab2	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab3	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab4	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab5	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab7	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab8	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab9	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab10	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab11	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab12	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	lab13	Type: MatFlowIndicator Description: Directional indicator for material flows in a diagram
	budget	Type: SimpleInformationLevel Description: Available budget
	externalResources	Type: SimpleInformationLevel Description: Available external resources
	clearMandate	Type: SimpleInformationLevel Description: Clear mandate assignment
	partiesInvolved	Type: SimpleInformationLevel Description: Number of parties involved
	decisionSpeed	Type: SimpleInformationLevel Description: Speed of decision making
	adHocDesignDecisions	Type: SimpleInformationLevel Description: Number of adhoc design decisions
	designQuality	Type: SimpleInformationLevel Description: Quality of software design
	swQuality	Type: SimpleInformationLevel Description: Software quality
	swAcceptance	Type: SimpleInformationLevel Description: Software acceptance
	swUsage	Type: SimpleInformationLevel Description: Software usage
	prodQuality	Type: SimpleInformationLevel Description: Quality of production
	prodEfficiency	Type: SimpleInformationLevel Description: Production efficiency
	costs	Type: SimpleInformationLevel Description: Costs
	costEffectiveness	Type: SimpleInformationLevel Description: Cost effectiveness
	changePressure	Type: SimpleInformationLevel Description: Need to change
	mgtKnowledgeLevel	Type: SimpleInformationLevel Description: Management knowledge level
	decisionQuality	Type: SimpleInformationLevel Description: Quality of decisions made
	swMgtOrgCapacity	Type: SimpleInformationLevel Description: Software management organization capacity
	employees	Type: SimpleInformationLevel Description: Number of employees
	swResistance	Type: SimpleInformationLevel Description: Resistance against software
	mgtQuality	Type: SimpleInformationLevel Description: Management quality
	r1	Type: ProportionalityDelayed Description: External resrouces >> clear mandate
	r2	Type: ProportionalityDelayed Description: External resources >> management knowledge level
	r3	Type: ProportionalityDelayed Description: External resources >> management quality
	r4	Type: ProportionalityDelayed Description: External resources >> software management organization capacity
	r5	Type: ProportionalityDelayed Description: Budget >> external resources
	r6	Type: ProportionalityDelayed Description: Clear mandate >> number of parties involved
	r7	Type: ProportionalityDelayed Description: Costs >> cost effectiveness
	r8	Type: ProportionalityDelayed Description: Decision quality >> design quality
	r9	Type: ProportionalityDelayed Description: Decision speed >> number of ad hoc design decisions
	r10	Type: ProportionalityDelayed Description: Design quality >> software quality
	r11	Type: ProportionalityDelayed Description: Management knowledge level >> decision quality
	r12	Type: ProportionalityDelayed Description: Management quality >> resistance against software
	r13	Type: ProportionalityDelayed Description: Management quality >> software usage
	r14	Type: ProportionalityDelayed Description: Change pressure >> budget
	r15	Type: ProportionalityDelayed Description: Number of ad hoc design decisions >> design quality
	r16	Type: ProportionalityDelayed Description: Number of employees >> costs
	r17	Type: ProportionalityDelayed Description: Number of employees >> resistance against software
	r18	Type: ProportionalityDelayed Description: Number of parties involved >> decision speed
	r19	Type: ProportionalityDelayed Description: Production efficiency >> costs
	r20	Type: ProportionalityDelayed Description: Production quality >> cost effectiveness
	r21	Type: ProportionalityDelayed Description: Software acceptance >> software usage
	r22	Type: ProportionalityDelayed Description: Number of parties involved >> design quality
	r23	Type: ProportionalityDelayed Description: Software management organization capacity >> software quality
	r24	Type: ProportionalityDelayed Description: Software quality >> software acceptance
	r25	Type: ProportionalityDelayed Description: Software usage >> number of employees
	r26	Type: ProportionalityDelayed Description: Software usage >> productive efficiency
	r27	Type: ProportionalityDelayed Description: Software usage >> production quality
	r28	Type: ProportionalityDelayed Description: Production quality >> change pressure
	r29	Type: ProportionalityDelayed Description: Software usage >> resistance against software
	r30	Type: ProportionalityDelayed Description: resistance against software >> software acceptance
	r31	Type: ProportionalityDelayed Description: Software management organization capacity >> number of employees
	c1	Type: SimpleControl Description: Directly influencing a stock to keep it at setpoint value
	c2	Type: SimpleControl Description: Directly influencing a stock to keep it at setpoint value
	c3	Type: SimpleControl Description: Directly influencing a stock to keep it at setpoint value
	c4	Type: SimpleControl Description: Directly influencing a stock to keep it at setpoint value
	usage	Type: Performance Description: Evaluation of a stock on a [0,1] scale
	quality	Type: Performance Description: Evaluation of a stock on a [0,1] scale
	cost	Type: Performance Description: Evaluation of a stock on a [0,1] scale
	totalPerformance	Type: AccumulatedPerformance Description: Mean weighted-average performance from startTime to currentTime

Revisions

Added in v2.0.0.