PERD Client Library for Aggregating Siloed Petroleum Data, Enabling Elastic On-Demand Simulation Workspaces, and Reducing Foreign Software Dependence in Nigerian Oilfield Operations

Abstract Nigeria's oil and gas sector continues to face a persistent digital bottleneck: valuable subsurface and production data remain difficult to access, costly to acquire, and tightly constrained by proprietary controls, while advanced simulation and analytics ecosystems are still largely dependent on imported platforms. This limits local algorithmic research, slows innovation, and widens the gap between petroleum data availability and petroleum computation. This study presents the design and evaluation of the PERD client library, a Python-based access and execution interface developed as a core component of the Petroleum Engineering Research and Development (PERD - perd.app) platform to enable secure, governed, and computation-ready interaction with sensitive petroleum datasets. The client library is positioned not merely as a software convenience layer, but as an enabling front end for a broader cloud-native research infrastructure. Within the PERD architecture, it supports credential management, authenticated access, workspace orchestration, websocket session management, dynamic workflow invocation, and live visualization of computational outputs. A key innovation is its metadata-driven runtime workflow model, in which callable engineering functions are generated dynamically from deployment manifests rather than hard-coded into a static software development kit. This approach creates a flexible interface capable of adapting rapidly to newly deployed simulation and analytics workflows while preserving an intuitive user experience for high-performance engineering tasks. Results demonstrate that the system can provision new workspaces, reconnect efficiently to existing environments, sustain long-lived bidirectional workflow sessions, and support non-blocking visualization that continues updating independently of the originating execution path. In a stress test involving simultaneous long-running training jobs, the platform maintained streamed interaction for approximately one hour, confirming its suitability for compute-intensive research scenarios that exceed the temporal limits of conventional web request architectures. Visualization tests further showed that multiple live computational outputs can remain responsive in parallel, supporting the multi-job cadence typical of advanced modelling and AI experimentation. Taken together, these findings position the PERD client library as a promising digital instrument for narrowing the long-standing divide between petroleum data access and petroleum computation in Nigeria. By coupling secure governance, extensible workflow execution, and live analytical interaction within a single programmable surface, the platform offers a credible pathway toward faster local research, stronger indigenous technical capacity, and more economically efficient development of petroleum simulation and AI tools. The work suggests that PERD can evolve beyond a data-access solution into a foundational computational marketplace for locally driven innovation in the energy sector.