BIM Information Manager — Expert Program | Module 2: Interface Management & ICDs (Lead

Appointed Party PoV)

CHAPTER 2.5
Systems-Integration Interfaces (Signalling ↔ Power ↔ Comms ↔
Civil)
General-first teaching · Applied-to-AMIA segment · Verdict: Expert (Leaner Gauntlet: 2 Viva + 5 Red-
Team)

How to read this chapter

This chapter carries the highest concept density in Module 2 and includes four
learner doubts that materially sharpened the taught material — all captured below as
corrections/enrichments, not footnotes.
● correction · ● learner enrichment · ▲ tracked pattern

Part A — Body
2.5.1 Why systems interfaces are a different animal
Most interfaces (2.1–2.4) resolve by geometry and a document. Systems interfaces are
different because the thing crossing the boundary is BEHAVIOUR, not just geometry or a
data field — one system's real-time performance depends on the other's, continuously,
while both are running.
Property What it means Why it matters
Real-time coupling Continuous live flow during A design that works on paper can
operation, not a one-time fail under live timing/load
exchange
Safety-criticality Many sit inside a safety case Errors can cause a hazard, not
(signalling especially) just delay
Emergent behaviour Combined system behaves in Can't verify by inspecting either
ways neither system does system alone — only by testing
alone together

2.5.2 The four-way coupling: six pairwise interfaces

Pair What crosses Why it's hard
Signalling↔Power Power quality/availability; Reliability depends on quality, not
signalling demand profile just supply; demand is dynamic,
not static
Signalling↔Comms The data network signalling Hard latency/availability
rides on requirements ordinary IT
networks don't meet
Signalling↔Civil Equipment room space, cable Civil is designed first & is
routes, clearances, mounting slow/costly to change late
Power↔Civil Substation space, duct banks, Heavy loads/footprint — late
earthing, structural loading equipment change can break civil
design
Power↔Comms EMC/EMI — induced A physics interface — faults show
interference via proximity as intermittent data errors in
operation
Comms↔Civil Cable route diversity, Diversity must be planned into
duct/tray provision layout from the start — hard to
retrofit

2.5.3 EMC/EMI as its own interface class

EMC governs whether equipment operates correctly in the presence of electromagnetic
disturbance from OTHER equipment nearby. The interface isn't ‘A sends B a signal’ —
it's ‘A's normal operation must not degrade B's normal operation, purely through
physical proximity.’ Resolved through physical segregation, screening/shielding, and
bonding/earthing — not a data spec or meeting agreement. Closure evidence is a site
EMC survey/test.

2.5.4 Redundancy and degraded-mode interfaces

Safety-critical systems often carry redundancy (2-out-of-3 voting, N+1 supply). This
creates the DEGRADED-MODE INTERFACE: what must happen at a boundary when one
side has PARTIALLY FAILED, not just in normal operation. An ICD for a redundant/safety-
critical coupling must carry a NORMAL-MODE definition AND a DEGRADED-MODE
definition — two behaviours, one interface. (Deep RAMS analysis deferred to the
dedicated RAMS & QA/QC track.)

2.5.5 Closure evidence gets heavier: IRS and SIT

• IRS (Interface Requirement Specification) — the technical sibling of the ICD: input
requirements (signal lists, protocols, timing budgets, tolerances) plus, once tested,
the recorded output/result data. ICD = governance wrapper (owner, dates, closure);
IRS = the engineering content it references.
• SIT (System Integration Test) — closure evidence is a live test of two systems
operating together, not a drawing. Heavier evidentiary bar, consistent with 2.5.1's
behavioural (not geometric) nature.

2.5.6 The LAP lens: systems integration workshops

The weekly ICM (2.4) is often too broad for the hardest four-way coupling. Expert
practice: convene DEDICATED systems-integration workshops for
signalling↔power↔comms↔civil specifically — smaller, technical, focused. The
workshop is where the engineering happens; the register/ICM is still where it's tracked
and governed. A workshop is not a second source of truth: it is the same governance
thread, disciplined to report back into the existing ICD/Interface ID before the next ICM.

2.5.7 Design-stage prevention vs SIT verification (corrected via learner

doubts)

● Correction — EMC prevention happens at design stage, not at SIT (Doubt 1)

SIT/site survey CANNOT predict EMC — it verifies it. Prevention happens earlier: EMC
segregation distances, screening and cable routing are calculated from known physics
at design stage and built into the civil design BEFORE anything is embedded. Once
equipment is fixed via bolted/embedded methods, civil relocation is effectively
foreclosed — the only remaining post-installation lever is cabling-level (shielding,
ferrite suppression, re-routing cable trays), not structural. This is precisely why EMC
must be calculated at design, not discovered at SIT.
The design calculation is not a perfect prediction — it includes a
MARGIN/TOLERANCE precisely because as-built installation always deviates
somewhat from the drawing. SIT verifies the margin held; if deviation ate through the
buffer, the fix is a compensating measure (screening/encasement/ferrite suppressor
via an approved method statement) — not automatically a full re-route.

● Correction — ICM/interface management is preventive IN STAGES, not uniformly

(Doubt 2)
At design (pre-freeze): prevents the WRONG requirement ever being built — fully
preventive, ~90% of the value.
Freeze point → installation: prevents UNCONTROLLED CHANGE — locks the agreed
spec.
SIT (post-installation): NO LONGER preventive of design error — it is detection &
containment. Its value is catching the failure before commissioning/revenue service,
and triggering the cheapest available fix (cabling/software/calibration) rather than
the most expensive (structural rework).
Correct statement: 'ICM at SIT stage is not preventive of the underlying error — it's
preventive of the WORSE outcome (finding the fault in passenger operation instead
of during testing).'

● Enrichment — the two-milestone ICD

Where closure evidence depends on post-installation SIT, an interface legitimately
stays Open/In-Progress for an extended duration between design-stage discovery
and site testing. This is correct behaviour, not a flaw — track it as ONE ICD carrying
TWO dated milestones: a near-term DESIGN FREEZE (input side locks: revised spec,
EMC calc, mitigation spec) and a long-term TEST CLOSURE (once SIT completes). Not
two interfaces — one ICD, two milestones.

Clarified — SIT vs NDT (Doubt 3)

NDT (Non-Destructive Testing): examines a PHYSICAL/MATERIAL property without
damaging the item (e.g. ultrasonic weld test) — answers 'is this physical thing
sound?' Applies to ONE component in isolation.
SIT: verifies BEHAVIOUR between two or more systems operating together —
answers 'do these systems work correctly together, in real time?' Different question,
different method, different stage. Both can apply to the same asset at different
points.

● Enrichment — mock-ups as budget-friendly prevention, with a real limit (Doubt 4)

Mock-ups are genuinely cost-effective and preventive for PHYSICAL/SPATIAL risk
(clearances, access, geometry) — use wherever feasible, before installation.
They do NOT substitute for a full SIT on real-time, safety-critical systems behaviour —
emergent behaviour (2.5.1) arises from the ACTUAL hardware/software combination,
which a mock-up cannot fully replicate. A mock-up-verified physical interface reduces
what the live SIT still needs to check, but doesn't eliminate the functional SIT on
safety-critical couplings.

Applied to AMIA APM

The four-way coupling is: L&T's signalling (CBTC, GoA4) ↔ L&T's traction power ↔
L&T's communications ↔ DAEP's Fixed-Facility civil contractor
(guideway/tunnel/station structures, explicitly outside L&T's systems scope). Every
signalling/power↔civil pairing is therefore ALSO a cross-contractor (L&T↔Fixed-
Facility) interface, not just cross-discipline. EMC segregation between power and
signalling/comms cabling sharing guideway/tunnel duct banks is a genuine physical
risk. Interview line: run the standard weekly ICM for general coordination, but stand
up dedicated systems-integration workshops for the signalling-power-comms
coupling, with EMC and SIT as named work-streams and IRS documents feeding the
governing ICDs.
Part B — Doubts (4 raised — all captured above as
corrections/enrichments)
• Doubt 1: EMC/SIT can't be predicted, only shielding is left post-installation given
fixed bolt/embed methods → corrected the Body: prevention is at design stage
(with margin); SIT verifies.
• Doubt 2: if IRS shows failure post-installation, what does ICM actually prevent, since
dismantling costs money? → corrected the Body: prevention is staged, SIT is
detection/containment not prevention of the original error.
• Doubt 3: is SIT the same as NDT? → clarified: different question (behaviour vs
material), different method, different stage.
• Doubt 4: are mock-ups budget-friendly for SIT? → enrichment: yes for
physical/spatial risk, not a substitute for functional SIT on safety-critical behaviour.

Part C — MCQ (Recall) — 30/30

Q1. What makes a systems-integration interface fundamentally different from general
coordination?
A) Two systems must physically fit in the same space
B) One system's behaviour depends on another's real-time behaviour while both are
operating ✓
C) Two systems share the same contractor
D) Two systems use the same software platform
Feedback: 10/10.
Q2. What is appropriate closure evidence for most systems-integration interfaces,
beyond a drawing?
E) A coordinated drawing only
F) A System Integration Test (SIT), verifying real, live behaviour ✓
G) A verbal ICM agreement
H) The IRS alone, with no physical test
Feedback: 10/10 — and correctly connected to Doubt 2: the real closure package is SIT
(the test) producing the IRS (the record), together, not either alone.
Q3. Per the corrected understanding, when does EMC prevention actually happen?
I) Mainly at SIT stage, by testing installed cabling
J) Mainly at design stage, via calculated segregation/screening built into the civil design;
SIT/survey mainly VERIFIES it ✓
K) EMC is not a real concern on rail/transit projects
L) Only by relocating embedded equipment after installation
Feedback: 10/10.
Part D — Scenario (Application) — 8/10, first pass
Prompt: Late in design, traction power confirms backup supply sustains full load for
only 90 seconds — shorter than assumed — affecting signalling, with an EMC concern
from backup-path rerouting through a shared duct bank with comms. Identify engaged
couplings, the new ICD content needed, closure evidence, and why this is still
preventive.
Evaluation — 8/10
Right: correctly identified the degraded-mode requirement for the ICD unprompted
(power to specify degraded behaviour + EMC generated; comms to verify via SIT; civil
correctly placed as Consulted for a routing option). Closure evidence was thorough —
revised spec/cabling layout, IRS with SIT sequence, ferrite suppressor detail placed in
the spec book (not the drawing) to avoid over-specifying, and main-vs-fault-condition
data comparison.
● Minor: described the coupling as a CHAIN (SIG→TPS→COMM→CIV) rather than
three separate pairwise couplings converging on one root cause (SIG↔TPS primary;
TPS↔COMM EMC; TPS↔CIV routing). Comms and civil are not downstream of each
other.
● Part 4 under-claimed the learner's OWN Doubt-2 correction: framed as 'less bad
than a risky fallback' (contained-outcome framing) rather than recognising this is
squarely a DESIGN-STAGE catch — fully preventive, the good case, not merely the
smaller blast radius.

● Learner's own doubt, confirmed as a valid enrichment

Learner asked: if closure depends on SIT (which needs installed equipment), doesn't
the interface stay open from FD-stage discovery until site SIT? Confirmed correct —
and formalised as the TWO-MILESTONE ICD (design freeze near-term; test closure
long-term) captured in 2.5.7 above.

Part E — Rapid-Fire (Terminology) — 6/8 → re-cleared

# Definition Answer Verdict
1 Combined-system failure mode invisible in “Degraded mode” ✗→
either system alone emergent
behaviour
2 Interface governed by segregation/screening, “4-way coupling” ○ → EMC
not a data spec specifically
3 ICD's technical sibling — inputs + recorded IRS ✓
outputs
4 Two behaviours a safety-critical ICD must define Normal + ✓
degraded mode
5 Live test of two systems operating together SIT ✓
6 Test of a physical/material property, NDT ✓
# Definition Answer Verdict
undamaged
7 Dedicated forum for the hardest 4-way coupling Systems- ✓
integration
workshop
8 Stage where EMC prevention actually happens Design / final ✓
design
6/8 first pass. Isolated re-run (only #1, #2): ‘emergent behaviour’ and ‘Electromagnetic
Compatibility (EMC)’ — both correct. Rapid-Fire CLEARED.
▲ Note on #1
Not the usual naming-under-load pattern — degraded mode and emergent
behaviour are two DIFFERENT, adjacent 2.5 concepts (a design requirement you write
into the ICD, vs. a property of combined systems explaining why the discipline is
hard). Keep them cleanly separated.
Part F — Viva Cross-Examination (2 challenges) — 8 · 8
Challenge 1 — “As-built never matches the drawing, so isn't SIT the REAL
prevention?”
Panel: Design calculations rely on assumptions that installation deviation breaks —
doesn't that make SIT the true test of reality?
Answer (summary): First pass repeated the value of design-stage EMC without
engaging the as-built-deviation mechanism — 6/10. Re-run (concrete 300mm-vs-
250mm clash) answered directly: spec anticipates deviation via a pre-defined remedy
(EM screen/barrier, same logic as concrete encasement for wet-services clashes),
governed via variation spec or a formal method statement where not pre-covered —
8/10 CLEARS.
Evaluation
Design-stage EMC calculation includes a MARGIN, not a knife-edge minimum — built
with tolerance because installation deviation is expected, not a surprise. SIT/survey
verifies the margin held; if deviation ate through it, the fix is a compensating measure
(screening, encasement, ferrite suppression), which is far cheaper than no
segregation having been designed at all.

Challenge 2 — “Splitting governance into workshops = a second source of truth.”

Panel: Why should a separate systems-integration workshop be safer rather than just
another place for information to get lost?
Answer (summary): 8/10 CLEARS on first pass. Correctly reframed: a workshop is the
SAME ICM governance devoted to one issue, not a parallel record — outcomes (mock-
up, simulation, or on-site TS survey) all recorded in the SAME ICD under the SAME
Interface ID; closure evidence in the same CDE container with a transmittal referencing
the ID.
Evaluation
A systems-integration workshop is an ICM session with one item on the agenda and
more technical depth in the room — same governance, same register, same ICD. The
safeguard against fragmentation is that the workshop's output must be logged
against the existing Interface ID before the next ICM — not an automatic guarantee,
but a disciplined reporting rule.
Part G — Red-Team Audit (5 items) — 9/10, all 5 found
Exhibit F: a flawed systems-integration procedure. All five planted defects found and
corrected — each fix drawing directly on this chapter's Viva arguments.
# Planted defect Correct position
1 ‘No design-stage EMC calc needed, SIT Primary calculation IS required at design
confirms it, avoids duplication’ stage, with safety factor/tolerance built
in; SIT confirms, doesn't replace it
2 ‘ICD only needs the normal operating Both safety-critical systems require
condition’ normal AND degraded-mode verification,
during interfacing and at SIT
3 ‘Deviation = non-compliant, must fully Non-compliant status, but not automatic
re-route civil ductwork’ full re-route — an approved method
statement with
screening/encasement/ferrite suppressor
(spec’d) as a compensating measure,
approved prior to installation
4 ‘SIT results go straight into the register SIT results recorded in the IRS FIRST (raw
status field, IRS unnecessary’ report attached); synopsis + IRS ID
referenced in the register; no SIT-based
verification accepted without an IRS
5 ‘Workshop outcomes wait for the next Workshop MoM as process is fine, but
monthly steering committee’ ICD/IR MUST be updated and shared with
stakeholders immediately after the
workshop — not held for the next
scheduled forum
Genuine transfer, Viva → Red-Team
#1 states the Viva-1 margin/tolerance defence as a governance rule. #3 applies the
Viva-1 'compensating measure, not automatic failure' precisely, with the governance
gate (approval) intact. #5 supplies exactly the reporting-discipline safeguard the Viva-
2 evaluation asked for — all three concepts moved from adversarial defence into
audit-ready rule-writing within the same session.
Part H — Record: Pattern, Key Terms, Glossary
▲ Tracked pattern
This chapter's doubts were the strongest evidence yet that the learning loop is closing
in real time: four doubts became correct, load-bearing arguments (design-stage margin,
staged prevention, two-milestone ICDs), successfully defended under adversarial Viva
pressure minutes later, then applied unprompted in the Red-Team audit. The residual
naming-under-load thread persists only lightly (Rapid-Fire #2), and one genuinely new,
non-pattern confusion appeared (#1: degraded mode vs emergent behaviour —
adjacent concepts, not a naming slip).

“Use This Term When…”

When you mean… Say this term
why testing together reveals faults Emergent behaviour
inspection can't
segregation/screening-governed interface EMC (Electromagnetic Compatibility)
design isn't a perfect prediction, it has Design margin / tolerance
headroom
why SIT-stage catches aren't fully Detection & containment, not design
preventive prevention
an interface open a long time awaiting a Two-milestone ICD (design freeze + test
test closure)
deviation found on site, not an automatic Compensating measure via approved
failure method statement

Terms introduced in this chapter (feeds the Module 2 glossary)

Term Definition
Emergent behaviour A combined-system behaviour neither system exhibits alone;
only detectable by testing together
Four-way coupling Signalling↔Power↔Comms↔Civil — six pairwise
interfaces from n(n−1)/2 with n=4
EMC / EMI Electromagnetic Compatibility/Interference — governed by
segregation, screening, bonding/earthing, not a data spec
Degraded-mode interface What must happen at a boundary when one side has partially
failed — must be explicitly defined alongside normal mode
IRS Interface Requirement Specification — technical input
requirements + recorded SIT outputs; the ICD's technical
sibling
SIT System Integration Test — live verification of behaviour
between two+ systems operating together
Term Definition
NDT Non-Destructive Testing — physical/material soundness of
one component, undamaged
Systems-integration Dedicated LAP-convened forum for the hardest coupling —
workshop same governance as the ICM, more technical depth
Design margin / tolerance Built-in buffer in a design calculation accounting for expected
as-built deviation
Two-milestone ICD One ICD carrying a near-term design-freeze milestone and a
long-term test-closure milestone

Score summary
Layer Score Status
MCQ 30/30 ✓ Cleared
Scenario 8/10 ✓ Cleared (first pass)
Rapid-Fire 6/8 → re-cleared ✓ Cleared
Viva 8·8 ✓ Cleared (2 challenges)
Red-Team 9/10 ✓ Cleared (5 items)
Verdict: Chapter 2.5 cleared at expert standard.