CONTEXTUAL KNOWLEDGE COORDINATIONIN AIR TRAFFIC CONTROL OPERATIONSMeteorological Field Studies at CRNA/N (Athis Mons) and Orly Tower

Research Problem. Some complexity of socio-technical systemslies in the range of performance varieties required to sustain usefulfunctions. We address operator performances in the ATM system whenfaced by meteorological (MET) events. We focus on the exchange of

Research Approach- A Resilience Engineering framework (Hollnagel, E., Woods, D., &Leveson, 2006) provides the systemic framework to qualify MET as externaldisturbances on the ATM system.

We hypothesize that the coordinative performances of controllers in ATMvaries along the time horizon (leading to a goal) by shifting linearly between 3boundary-spanning modes: a) syntactic, b) semantic and c) pragmatic. Theperformance mode at a point in time is influenced first by the context of work,y g ( ) g

expert knowledge underlying the coordination of controller teams andsee how new working strategies are developed during MET disturbances.

y- A Computer Supported Cooperative Work (CSCW) framework providesthe qualification of ATM teams as Communities of Practice (CoP). Eachcommunity coordinates knowledge across its specialised knowledgeboundaries via coordinative artifacts.

Long <<< Temporal Horizon >>> Short

Supervisory Control Mode

Strategic Tactical Opportunistic Scrambled

which is qualified by Hollnagel (1993) as disturbances from theenvironment. In our case, such disturbances are CBs or MET events in ATM(Figure 2). Second, performance modes also depend on the experience,expectations and resource availability of the operators. Finally, creativeworking strategies increase when operators are in an ‘exploratory control’mode of interaction.

Weekly and Annual Delay Distribution 2004-2007

ATM-MET Interaction, Conceptual View(Joyekurun, Wong, & Amaldi, 2008)

Control Mode (Hollnagel, 1993)Knowledge Coordination Mode (Carlile, 2002)

Syntactic Semantic Pragmatic

Contextual Control and Coordination Modes

Methodology. The methodological approach isgrounded in ethnography. The methods choices wereadapted to the ATM context and developed using a strategybounded by procedure, ethical and time constraints.

Methods were varied across situations, based on anassessment of controller workload. Following this strategy:-

ResultsSyntactic Mode - Syntactic mode ofboundary spanning addresses frequentlyoccurring performances and is effectivelymediated by ATM procedures and standards.

E g Phraseology – Standard Aviation Phraseology is one of the

Semantic Mode - Semantic mode of boundaryspanning occurs when procedures are limitingand need to be circumvented – this is frequentlyreferred as deviations from standardperformances.

Pragmatic Mode - Pragmatic mode of boundary spanning occurs when the standardperformances for resolving a situation are highly inadequate or cannot be satisfactorilyre-adapted – operators often refer to the solution as a matter of experience.

E.g. Flow Regulation – The experience of controller with different flow re-adjustments allows an expectation of traffic patterns andworkload. This is not part of formal training and only develops as a controller becomes very experienced after moving acrossg gy

- Perceived high workload periods leads to application ofparticipant observation, and- Low workload periods incurred the application of shadow-mode and expert interviews.

Perceived workload was estimated from a number ofcontextual factors such as:the “hum” in control room (low vs. high); estimated phonecalls per minute; perceived attention of controllers to radarscreens (focused vs. dispersed); involvement of supervisorwith team members; break allowances (on time delayed

E.g. Phraseology Standard Aviation Phraseology is one of the more salient procedural instances of coordination among pilots and controllers. It was previously observed to be much more strictly used at Orly Tower than at CRNA/N [REF 3].

[REF 3; ¶ 90]OBS: Phraseology is hardly ever followed. If a conversation followed phraseology but was not understood by either the pilot or the controller (asking for 'say again'), then exchanged pursues in natural language before phraseology is used again to punctuate the order given/taken. [Delivery mode varies with level of uncertainty].

E.g. Vertical Separation, Flow Regulation – These are standard methods applied when CBs are incident on traffic flow. Controllers assume the vertical separation of traffic [REF 24] and flow can be regulated by feeding back estimated sector capacity changes to

E.g. Capacity Estimation - The calculation of capacity is a highly situatedactivity. The workload incurred by controllers for handling this increasedcoordination activity forms an additional constraint on the system. In [REF3], the supervisor takes into account the perceived workload of controllersas a means of informing his decision.

[REF 31; ¶339] C: Thunderstorms are still the prime factor for us. Capacity can easily drop from 38 to 25. The problem with fog is that when it has not been forecasted, you accept 38 cap (capacity) and then you have to manage them - finding parking spots and so on is hard. As for snow, it is notable although the least significant - the issue is that given it is an infrequent event, the response teams are not always prepared to quickly handle it.

E.g. MET Information Correlation, Objectivity – The provision of weatherforecast and real-time MET products are included in control centre

different roles in the control centre.

[REF 14; ¶159]OBS: A controller seems upset and complains of inadequate flow management (fr: c'est quoi ce regul, la?).Q: What happened with the flow management?R: They are being over reactive again! Why did they close off so much? Now we are going to get a surge in a few hours time. [..]Q: Why did they do that?R: They want to reduce capacity because of the CB. Now they closed it off too much and we're gonna get all the load at 11:00.Q: What would have been preferable for you guys?R: I would rather have a moderate flow constantly than spikes - makes sense. You do not want all (CWP) positions open and only a few aircraft moving; while you wait for the peak to catch up in a few hours time.

Conclusion. The coordination modes use for boundary spanning atCRNA/N and Orly Tower seems to follow three different modes: 1) syntactic 2)with team members; break allowances (on-time, delayed,

interrupted, uninterrupted); attendance to CWP bank(present, short absence, long absence); readiness ofcontrollers to speak to the practitioner (deferred response,immediate response – short vs. long, unacknowledgedresponse, non-initiated response).

regulated by feeding back estimated sector capacity changes to CFMU. While a capacity calculation tool exists, it is not used often by the control room supervisor [REF 1]. Instead, a more informal technique is used involving the controller team leaders and a pragmatic evaluation of CB impacts on the work of controllers .

[REF 1; ¶ 64]Q: How do you reach the capacity value then?R: Capacity is reached mainly through experience. It can also be informed by a calculation tool which is not used often. I'll speak to the team leaders and take a look around the room. Then if there are non-nominal events such as CB I can re-adjust the capacity.

forecast and real-time MET products are included in control centreprocedures. The applicability of such products to controllers’ decisions isnot guided by procedures. While procedures are specific, the decisionwhether a procedure is applicable is variable and liable to deviations.

[REF 11; ¶130] Q: Do you know where the CB is at this moment?R: Seems to be around here [points to screen] because they (pilots) are asking for headings. Also it was around here [points to the lower left corner of the screen] and with the North-East wind, it should be here roughly.

CRNA/N and Orly Tower seems to follow three different modes: 1) syntactic, 2)semantic, and 3) pragmatic. Each mode can be activated sequentially as thetemporal horizon to the accomplishment of a goal decreases. ATM procedures(syntactic spanning) are seen to address frequent and salient work patterns;deviations (semantic spanning) are required to address moderately occurringpatterns; while experience (pragmatic spanning) addresses exceptional cases.Future work attempts to map the coordination modes to boundary artifact design.

Ronish Joyekurun†‡, Paola Amaldi† & William Wong†

Ronish.Joyekurun.ext@eurocontrol.int; {P.Amaldi-Trillo, W.Wong}@mdx.ac.uk†Interaction Design Centre, Middlesex University, London NW4 4BT, UK.‡EUROCONTROL Experimental Centre, F-91222 Bretigny Sur Orge, France.

REFERENCESHollnagel, E., Woods, D., & Leveson, N. (Eds.). (2006). Resilience Engineering: Concepts and Precepts. England: Ashgate.Hollnagel, E. (1993). Human Reliability Analysis: Context and Control. London: Academic Press.Carlile, P. (2002). A Pragmatic View of Knowledge and Boundaries: Boundary Objects in New Product Development. Organization Science, 13(4), 442-455.Joyekurun, R., Amaldi, P., & Wong, W. (2007). Weather Hazards in ATM: Designing for Resilient Operations. Paper presented at the European Conference on Cognitive Ergonomics. Joyekurun, R., Wong, W., & Amaldi, P. (2008). Responding to Uncertainty on Approach in Hazardous Situations. Paper presented at the ICRAT 2008: International Conference on Research in Air Transportation, 2008.,