Engineering partner for SaaS & platform companies.
Book a Strategy Call →Have a smaller task? Submit a quick request →Copyright 2025© Peregrine-it.com | All Rights Reserved
How we built a manufacturing ERP system spanning 8 factories and 50K SKUs — with multi-plant sync, BOM management, shop floor data capture, and 30% inventory reduction.
Multi-Plant ERP · BOM · Supply Chain
A mid-size manufacturer with 8 plants across three states was running on a mix of spreadsheets, legacy MRP software, and SAP for finance. Each plant had its own way of tracking production — some used paper travelers, others Excel. Inventory counts were done monthly and reconciliation took weeks.
They wanted a unified ERP that could sync production and inventory across all plants in real time, manage complex bills of materials, capture shop floor data via tablets, and integrate with SAP for orders and financials.
Each plant operated as a silo. Transfer orders between plants took days to reflect. Corporate had no real-time view of total inventory or production status across the network.
Bills of materials were managed in spreadsheets. Multi-level BOMs with dozens of components were error-prone. Changes propagated manually — version drift between plants caused production issues.
Operators recorded production and downtime on paper. Data entry happened at end of shift — by then accuracy suffered. No real-time visibility into line utilization or quality metrics.
Purchasing relied on reorder points in spreadsheets. Lead times and supplier performance were tracked manually. Stockouts and overstock were common — no demand forecasting.
We built a centralized manufacturing ERP with event-driven architecture. Shop floor tablets publish production events via MQTT. A sync service aggregates data and updates inventory in real time. BOMs are versioned and shared across plants. SAP integration handles sales orders and financial posting.
The MQTT bridge was designed for unreliable shop floor networks. Tablets buffer events locally and retry until acknowledged. We use QoS 1 to ensure at-least-once delivery. Duplicate events are deduplicated server-side using event IDs.
Mapped production workflows across 8 plants. Audited existing BOM structure and SAP integration points. Designed multi-plant schema with plant-specific inventory buckets.
Built inventory management with multi-plant sync. Implemented versioned BOM engine with multi-level explosion. Developed work order and production planning modules. Integrated with SAP for order and material data.
Deployed MQTT broker and built ingestion pipeline. Developed tablet app for production entry and downtime logging. Built real-time dashboards for plant and corporate views.
Implemented demand forecasting and reorder logic. Built supplier performance tracking. Phased rollout — 2 pilot plants first, then 6 more over 8 weeks. Trained users and tuned workflows.
Manufacturing ERP complexity scales with plants and SKUs. The multi-plant sync had to handle conflicting updates — we used optimistic locking and event sourcing for inventory movements. Transfer orders between plants are treated as first-class transactions with atomic commit.
MQTT was the right choice for shop floor data. Wi-Fi in plants is often spotty — tablets needed to work offline and replay when reconnected. We designed a thin client that buffers locally and syncs when network is available. Operators don't notice connectivity issues.
SAP integration required careful mapping. Material masters, plants, and storage locations had to align. We built an abstraction layer so our schema could evolve independently — the SAP adapter handles translation. Bi-directional sync needed conflict resolution — we treat SAP as source of truth for orders, our system for production and inventory.
We help manufacturers build production-grade ERP systems with real-time visibility. Let's talk about your architecture.