Malware and forensic analyses of embedded cyber-physical systems are tedious, manual processes that testbeds are commonly not designed to support. Additionally, attesting the physics impact of embedded cyber-physical system malware has no formal methodologies and is currently an art. This chapter describes a novel testbed design methodology that integrates virtualized embedded industrial control systems and physics simulators, thereby supporting malware and forensic analyses of embedded cyber-physical systems without risks. Unlike existing hardware-based testbeds, the resulting soft industrial control system testbeds are portable, distributable and expandable by design. However, embedded system virtualization is non-trivial, especially at the firmware level, and solutions vary widely depending on the embedded system architectures and operating systems. This chapter discusses how the proposed methodology overcomes the challenges to virtualizing embedded systems and explores the benefits via a proof-of-concept implementation involving a Siemens MJ-XL variable step voltage regulator control panel.